Вы находитесь на странице: 1из 108

Re-thinking synonymy:

semantic sameness and


similarity in languages and their
description
Book of Abstracts

28.-30.10.2010, Helsinki
Table of Contents
LIST OF PARTICIPANTS: ___________________________________________________________________________ 1

INVITED SPEAKERS _______________________________________________________________________________ 3

GEERAERTS DIRK_________________________________________________________________________________ 3
HASPELMATH MARTIN _____________________________________________________________________________ 4
LEVIN BETH ____________________________________________________________________________________ 5

SCIENTIFIC COMMITTEE___________________________________________________________________________ 6

ORGANIZING COMMITTEE ________________________________________________________________________ 6

WORKSHOP ON COMPUTATIONAL APPROACHES TO SYNONYMY _________________________________________ 7

GRAEME HIRST, KENTARO INUI AND STEDE MANFRED ________________________________________________________ 7

ABSTRACTS _____________________________________________________________________________________ 8

ANDERSSON MARTA ______________________________________________________________________________ 8


ANISHCHANKA ALENA, SPEELMAN DIRK AND GEERAERTS DIRK __________________________________________________ 9
ARPPE ANTTI AND DIVJAK DAGMAR ___________________________________________________________________ 11
BACKUS AD AND MOS MARIA _______________________________________________________________________ 13
BORIN LARS AND FORSBERG MARKUS__________________________________________________________________ 15
CALUDE ANDREEA AND PAGEL MARK __________________________________________________________________ 17
CAPPELLE BERT AND DESUTTER GERT __________________________________________________________________ 18
COPESTAKE ANN AND HERBELOT AURÉLIC _______________________________________________________________ 20
DALMAS MARTINE AND DOBROVOL’SKIJ DMITRIJ __________________________________________________________ 22
DESHORS SANDRA C. _____________________________________________________________________________ 24
DEVOS MAUD _________________________________________________________________________________ 26
FAULHABER SUSEN ______________________________________________________________________________ 28
GAILLARD BENOÎT, NAVARRO EMMANUEL AND GAUME BRUNO ________________________________________________ 30
GLYNN DYLAN _________________________________________________________________________________ 31
GLYNN DYLAN AND LEVSHINA NATALIA _________________________________________________________________ 33
GOROKHOVA SVETLANA ___________________________________________________________________________ 35
HALDER GUIDO ________________________________________________________________________________ 37
HUUMO TUOMAS AND LEHISMETS KERSTEN _____________________________________________________________ 39
HUUMO TUOMAS AND OJUTKANGAS KRISTA _____________________________________________________________ 41
HÄUSLER SABINE _______________________________________________________________________________ 43
JÜRINE ANNI __________________________________________________________________________________ 44
KLAVAN JANE __________________________________________________________________________________ 46
KHOLODILOVA MARIA ____________________________________________________________________________ 48
KHOUTYZ IRINA ________________________________________________________________________________ 50
LURAGHI SILVIA ________________________________________________________________________________ 51
LIU DILIN _____________________________________________________________________________________ 53
MARTÍ SOLANO RAMÓN AND RALUCA NITA______________________________________________________________ 56
MATUSCHEK MICHAEL AND GUREVYCH IRYNA ____________________________________________________________ 58
MONTAZERI NILOOFAR AND HOBBS JERRY_______________________________________________________________ 60
MULLI JUHA___________________________________________________________________________________ 62
NACEY SUSAN AND EGAN THOMAS ___________________________________________________________________ 64
NISSILÄ NIINA AND PILKE NINA ______________________________________________________________________ 66
NEDJALKOV IGOR _______________________________________________________________________________ 68
O’CONNOR KATHLEEN AND CORTEEL CÉLINE _____________________________________________________________ 70
OVERSTEEGEN ELEONORE __________________________________________________________________________ 72
PALOLAHTI MARIA ______________________________________________________________________________ 73
PAULSEN GEDA ________________________________________________________________________________ 75
PÄIVIÖ PIIA ___________________________________________________________________________________ 76
RAKHILINA EKATERINA AND TRIBUSHININA ELENA __________________________________________________________ 77
RAUKKO JARNO ________________________________________________________________________________ 79
RINGBOM HÅKAN _______________________________________________________________________________ 81
ROBERT STÉPHANE ______________________________________________________________________________ 82
ROCHE CHRISTOPHE AND CALBERG-CHALLOT MARIE ________________________________________________________ 84
SAMEDOVA NEZRIN ______________________________________________________________________________ 86
SCHMEISER BENJAMIN ____________________________________________________________________________ 88
SIROLA-BELLIARD MAIJA __________________________________________________________________________ 90
SOARES DA SILVA AUGUSTO ________________________________________________________________________ 92
SUTROP URMAS ________________________________________________________________________________ 94
STEDE MANFRED _______________________________________________________________________________ 95
WANG TONG AND HIRST GRAEME ____________________________________________________________________ 96
WIEMER BJÖRN AND SOCKA ANNA ___________________________________________________________________ 98
VILKKI LIISA __________________________________________________________________________________ 100
VÄSTI KATJA _________________________________________________________________________________ 102
ZENNER ELINE, SPEELMAN DIRK AND GEERAERTS DIRK _____________________________________________________ 104
List of Participants:
Last name First name Affiliation
Andersson Marta Stockholm University
Anishchanka Alena Univeristy of Leuven (QLVL)
Arppe Antti University of Helsinki & Alberta
Backus Ad Tilburg University
Borin Lars University of Gothenburg
Calberg-Challot Marie Onomia - Savoie Technoloac
Calude Andreea University of Reading
Cappelle Bert Univesity College Ghent
Copestake Ann Computer Laboratory, University of Cambridge
Corteel Céline Université d'Artois
Dalmas Martine Université Paris-Sorbonne
Deshors Sandra C. University of Sussex
Desutter Gert University College Ghent
Devos Maud Royal Museum for Central Africa
Divjak Dagmar University of Sheffield
Dobrovol’skij Dmitrij Russian Academy of Sciences
Egan Thomas Hedmark Univsersity College
Faulhaber Susen Friedrich-Alexander-University Erlangen-Nuremberg
Forsberg Markus University of Gothenburg
Gaillard Benoît CLLE-ERSS, Univiversity Toulouse 2
Gaume Bruno CLLE-ERSS, Univiversity Toulouse 2
Geeraerts Dirk University of Leuven
Glynn Dylan University of Lund
Gorokhova Svetlana St Petersburg State University
Gurevych Iryna Technische Universität Darmstadt
Halder Guido The University of Texas at Austin
Herbelot Arélic Computer Laboratory, University of Cambridge
Hirst Graeme University of Toronto
Hobbs Jerry University of Southern California
Huumo Tuomas University of Tartu
Häusler Sabine Martin-Luther-Universität Halle-Wittenberg
Jürine Anni University of Tartu
Kholodilova Maria Saint-Petersburg State University
Khoutyz Irina Kuban State University
Klavan Jane University of Tartu
Lehismets Kersten University of Tartu
Levshina Natalia University of Leuven
Liu Dilin The University of Alabama
Luraghi Silvia Università di Pavia
Last name First name Affiliation

1
Marti Solano Ramon University of Limoges
Matuschek Michael Technische Universität Darmstadt
Miestamo Matti Helsinki Collegium for Advanced Studies
Montazeri Niloofar University of Southern California
Mos Maria Tilburg University
Mulli Juha University of Eastern Finland at Joensuu
Murphy Lynne University of Sussex
Nacey Susan Hedmark Univsersity College
Navarro Emmanuel IRIT, Univiversity Toulouse 3
Nedjalkov Igor St-Petersburg State University
Nissilä Niina University of Vaasa
O'Connor Kathleen Université de Lille 3
Ojutkangas Krista University of Turku
Oversteegen Eleonore University of Tilburg
Pagel Markus University of Reading
Palolahti Maria University of Helsinki
Paulsen Geda Åbo Akademi University
Pilke Nina University of Vaasa
Päiviö Piia University of Toronto
Rakhilina Ekaterina Russian Academy of Sciences
Raluca Nita University of Nantes
Raukko Jarno University of Helsinki
Ringbom Håkan Åbo Akademi University
Robert Stéphane CNRS-LLACAN, Paris
Roche Christophe University of Savoie
Samedova Nezrin Azerbaijan University of Languages
Schmeiser Benjamin Illinois State University
Sirola-Belliard Maija University of Tampere
Soares da Silva Augusto Catholic University of Portugal
Soares da Silva Augusto Catholic University of Portugal
Socka Anna University of Gdansk
Speelman Dirk University of Leuven
Stede Manfred University of Potsdam
Sutrop Urmas Institute of Estonian Language, Tallinn, University of Tartu
Tribushinina Elena University of Antwerp
Wakayama Masayuki Aichi Shukutoku University
Wang Tong University of Toronto
Wiemer Björn Johannes-Gutenberg-Universität
Vilkki Liisa University of Helsinki
Västi Katja University of Oulu & Helsinki
Zenner Eline University of Leuven

2
Invited Speakers
Geeraerts Dirk
University of Leuven

Criteria for meaning discrimination and the identification of synonymy

In this talk, I will attempt to clarify the notion of synonymy against the background of the
theoretical and methodological developments in contemporary (lexical) semantics. My starting
point is the observation that the problem of synonymy and the problem of polysemy are essentially
the same: if we have valid criteria for establishing whether we are dealing with one or more
separate readings of a single item or construction, then the same criteria can be used for establishing
whether two items or constructions illustrate one or more meanings.
It follows, in general, that the difficulties with the criteria for establishing difference of
meaning that I analyzed in Geeraerts (1993) hold for synonymy as well as for polysemy. I will
illustrate the point by looking more closely at how the three main criteria distinguished in the 1993
article might be applied to the identification of synonymy.
More specifically, zooming in on the so-called 'definitional criterion' will allow me to go
beyond the 1993 framework by taking into account a number of important methodological
developments of the last two decades. I will have a look at a referential approach to lexical meaning
(as in Geeraerts, Grondelaers & Bakema 1994), a contextual feature approach (as in Speelman &
Geeraerts 2009), and a word-space based distributionalist approach (as in Heylen, Peirsman,
Geeraerts & Speelman 2008). For each of these approaches, I will try to indicate advantages and the
drawbacks, an exercise that will lead to the conclusion that there is as yet no single and generally
applicable method for establishing synonymy.
This conclusion will then be deepened by a critical look at the concept of synonymy itself: if
we re-think synonymy, could it be that the absence of a methodological consensus reflects not just
the weakness of linguistic methods, but a flaw in our conception of meaning ?

Bibiliography

Geeraerts, Dirk (1993). 'Vagueness's puzzles, polysemy's vagaries'. Cognitive Linguistics 4: 223-
272.
Geeraerts, Dirk, Stefan Grondelaers and Peter Bakema (1994). The Structure of Lexical Variation.
Meaning, Naming, and Context. Berlin/New York: Mouton de Gruyter.
Heylen, Kris, Yves Peirsman, Dirk Geeraerts and Dirk Speelman (2008). 'Modelling word
similarity: An evaluation of automatic synonymy extraction algorithms'. Proceedings of
the Sixth International Language Resources and Evaluation Marrakech: European
Language Resources Association. (Link: http://www.lrec conf.org/proceedings/lrec2008
/pdf/818_paper.pdf.)
Speelman, Dirk and Dirk Geeraerts (2009). 'Causes for causatives: the case of Dutch 'doen' and
'laten''. In Ted Sanders and Eve Sweetser (eds.), Causal Categories in Discourse and
Cognition 173-204. Berlin: Mouton de Gruyter.

3
Haspelmath Martin
Max-Planck Institute for Evolutionary Anthropology

The role of homosemasy (semantic equivalence) for constructing


comparative concepts in cross-linguistic studies

Linguists who specialize in the comparison of languages routinely make use of homosemasy,
i.e. semantic equivalence across languages, as a tertium comparationis. This is true not only for
lexical comparison (whether historical-comparative or typological), but also for most cases of
morphosyntactic comparison: purely form-based grammatical comparison is too difficult, because
forms vary too much. In fact, typologists have often claimed that typological comparison of
grammatical patterns is exclusively meaning-based, but this is too strong – in addition to semantic
similarity, typological comparative concepts often involve some formal similarity as well.
An important observation is that just as there is no true synonymy, there is no true
homosemasy: Meanings are often similar across languages, but rarely match perfectly. Cross-
linguistic semantic variation is rampant. This means that cross-linguistic comparison must involve a
set of special semantic concepts (which I call comparative concepts), which are not meanings of
particular language forms, but are created specifically for the purpose of language comparison. I
illustrate such comparative concepts from the World Loanword Database (Haspelmath & Tadmor
2009) and from their use in semantic maps (Haspelmath 2003). Unless one adopts an extreme
decompositional approach, the comparative concepts cannot be used for the description of
language-particular categories. Finally, I show that the same holds for morphosyntactic, formally
defined entities such as cases, relative clauses or converbs: Comparative concepts are separate from
descriptive categories, and language comparison and language description/analysis are more
independent of each other than is often thought (cf. Haspelmath 2009, 2010).

Bibliography

Haspelmath, Mark, & Uri Tadmor (eds.). 2009. World Loanword Database. Munich: Max Planck
Digital Library. http://wold.livingsources.org/.
Haspelmath, Martin. 2003. The geometry of grammatical meaning: Semantic maps and cross-
linguistic comparison. In Michael Tomasello (ed.), The new psychology of language:
Cognitive and functional approaches to language structure, vol. 2, 211–242. Mahwah,
NJ: Erlbaum.
Haspelmath, Martin. 2009. La typologie des langues pourquoi est-elle possible? Bulletin de la
Société de Linguistique de Paris 104. 17-38.
Haspelmath, Martin. 2010. Comparative concepts and descriptive categories in cross-linguistic
studies. Language 86(3).

4
Levin Beth
Stanford University

Synonymy and Arbitrariness in Linguistic Argumentation

Among the questions posed in the call for papers for the conference is "What does synonymy
(at any level/in any form) reveal about language?" A recurring answer is that synonymy --- or even
near-synonymy --- can be used to argue for arbitrariness in the form-function (or semantics-syntax)
mapping. Thus, researchers cite pairs such as "leaves" and "foliage", "mail" and "letters", or
"poems" and "poetry" to argue for the arbitrariness of the mass/count distinction (e.g., Chierchia
1998, Ware 1975). Similarly, pairs like English "blush" and its Italian translation equivalent
"arrossire" have been used to argue for the arbitrariness of the unaccusative/unergative verb
distinction (e.g., Rosen 1984) and pairs like alternating "sour" and transitive-only "embitter" or
alternating "shake" and intransitive-only "shiver" for the arbitrariness of the causative alternation
(e.g., Farsi 1974).
In this talk, I will revisit some of these examples and suggest that they do not actually support
arbitrariness in the form-function mapping for the associated phenomena. I will argue that these
pairs at best represent near-synonyms so that their members do differ in meaning and that this
difference in meaning is one that is critical to their difference in behavior. I will further suggest that
such pairs arise because meanings are construals of the world, so that even if in one instance
"leaves" and "foliage" might refer to the same entity -- the basis for the synonymy claim -- the two
words capture different representations of this entity. In fact, it is this point that underlies
Wierzbicka's well-known study of "oats" and "wheat" (1985). The conclusion, then, is that words
with similar meaning constitute a rich domain for studying the semantics-syntax mapping and for
developing finer-grained lexical semantic analyses.

5
Scientific committee
Arppe Antti (University of Helsinki)

Austin Peter (SOAS, London)

Creissels Denis (University of Lyon)

Divjak Dagmar (University of Sheffield)

Goldberg Adele (Princeton University)

Gries Stefan (UCSB)

Huumo Tuomas (University of Tartu)

Janda Laura (University of Tromsø)

Jantunen Jarmo (University of Oulu)

Luraghi Silvia (University of Pavia)

Rice Sally (University of Alberta)

Siewerska Anna (University of Lancaster)

Wälchli Bernhard (University of Berne)

Organizing committee
Arppe Antti (University of Helsinki)

Kittilä Seppo (University of Helsinki)

Kyröläinen Aki-Juhani (University of Turku)

Niemelä Maarit (University of Oulu)

Nikolaev Alexandre (University of Joensuu)

Rostila Jouni (University of Tampere)

Vartiainen Turo (University of Helsinki)

Visapää Laura (University of Helsinki)

6
Workshop on Computational Approaches to Synonymy
Graeme Hirst, Kentaro Inui and Stede Manfred
University of Toronto
Tohuku University
University of Potsdam

Many problems and applications in computational linguistics and natural language processing
implicitly invoke, in various forms, the concept of synonymy or identity of meaning. In one way or
another, they involve either determining identity (or non-identity) of meaning in different surface
forms or creating different surface forms for a single meaning.
For example, paraphrase recognition is an important component of the more-general problem
of recognizing textual entailment. Textual tailoring and personalization seeks to find the most
effective linguistic realization of a message for a particular user; automatically simplifying texts and
creating stylistic variations are special cases of this. Lexical choice in text generation tries to find
the best word for a given meaning and to discriminate it from other words that are close in meaning
but not synonymous in the context. Cross-lingual document retrieval and other cross-lingual
applications such as, in particular, machine translation conflate the ideas of synonymy and
translation equivalence.
But while there has been a large amount of research on computational methods for
determining degree of similarity in lexical meaning and for recognizing paraphrase, little attention
has been given to theoretical considerations of synonymy. Mostly, it is treated as a boolean property
(two words are or aren't in the same synset; two sentences are or aren't mutual entailments) with
little thought of any theoretical underpinning.
On the other hand, the real-world linguistic problems that natural language processing
addresses provide useful test cases for linguistic theories of synonymy, and the computational
methods developed are de facto theories of synonymy even if not intended as such.
This workshop will explore computational approaches to synonymy, with an emphasis on
explicating their implicit theoretical notions and their implications for linguistic theory.

7
Abstracts
Andersson Marta
Stockholm University

Synonymy of result adverbials

The purpose of this paper is to investigate semantic, syntactic, pragmatic, discourse and
textual properties of five one-word adverbials (so, then, thus, therefore, hence) marking the specific
function of the causal relation, i.e. result in written discourse of the BNC. These words are
considered roughly synonymous, yet we should be skeptical about their full synonymy, even though
the hypothesis is that they are all oriented towards a conclusion and cue a relationship of resultative
nature. Yet the semantic space of result is not evenly divided between them and each of them can be
understood to express another cognitive way of thinking about result, which is related to certain
pragmatic, semantic, distributional and, possibly, discourse-textual differences.
The paper will demonstrate and discuss these differences and try to answer the question why
English has all these different ways of licensing result relations. The words in focus will be
compared as members of an onomasiological set, which yields not only similarities between them,
but also certain differences, as some are more prototypical of result than others. Consequently, this
may lead to the questioning of the aptness of the label “result” (which even intuitively seems to be
very general) as describing the role of all the investigated words in managing information flow, text
structure and social interaction. Thus an investigation of the similarities and differences in the
behavior of so, therefore, hence etc. will not only be situated in the area of purely linguistic
exercises, but will also touch upon the area of human cognition and social needs for making such
distinctions.

8
Anishchanka Alena, Speelman Dirk and Geeraerts Dirk
University of Leuven
University of Leuven
University of Leuven

Co-extension and near-synonymy in color categorization

Research question – Continuing the line of research set out in Geeraerts (2006), Geeraerts et
al. (1994), Geeraerts and Speelman (2010), the study addresses the interaction between concept-
related semantic features and sociolectal parameters underlying lexical heterogeneity. By mapping
the referential range of color terms in internet product catalogs on the basis of their RGB values, we
can explore the conceptual and contextual factors that determine the extension of color categories of
different specificity and the patterns of their referential overlap. Specifically, we focus on the
phenomena of co-extension and near-synonymy as the major construal types in overlapping
categories.
Background – The analysis applies the categorization model proposed by MacLaury (1992,
1997) that specifically addresses the multiple instances of referential overlap and near-synonymy in
the color domain. Using vast empirical data, MacLaury identified four types of category construal -
inclusion, coextension, near-synonymy and complementation - all having the same underlying
cognitive mechanism of (dis)similarity emphasis. Focusing on co-extension as the most common
type of referential overlap, he also demonstrated that the asymmetries in this type of construal
license the use of near-synonymous names within the same referential range. With the growing
recognition of (dis)similarity as a basic categorization mechanism (Jameson and Alvarado 2003,
Langacker 2009, Taylor 2003), this model proves particularly insightful for the linguistic analysis
of (near)-synonymy. Specifically, it allows accounting for salience effects, non-discreteness and
multidimensionality of lexical meaning that had been revealed by cognitive linguistic studies and
that provide major challenges for the notion of synonymy traditionally formulated in terms of
‘mosaic’ semantic fields, semantic features, or truth-conditional equivalence (Geeraerts et al. 1994,
Geeraerts 2006, Divjak and Gries 2006, Glynn 2010).
Method and design – The study is based on an extensive corpus of color names and color
samples (over 65 000 observations) used by US manufacturers for online marketing in four product
categories (automobiles, clothing, make-up, and house paints). The available material provides
linguistic, sociolectal and referential parameters of color naming in advertising and is particularly
suited for an independent measurement of referential range of the color names using the
automatically retrieved RGB values of the color samples.
Building on MacLaury’s criteria for inclusion, coextension, and near-synonymy, we explore
semasiological and onomasiological variation in overlapping color categories, such as purple,
mauve, lavender, lilac. Starting from a semasiological perspective, we first map the referential
range of basic and non-basic color words in order to identify the semasiological structure of color
categories and the patterns of co-extension and near-synonymy in their referential overlap. The
second stage investigates whether the degree and the pattern of referential overlap significantly
correlate with conceptual factors (such as vagueness and prototypicality of the color concept,
entrenchment and complexity of the color name, specificity of the color name and color concept)
and contextual factors (product category and type, target consumer). Taking the onomasiological
perspective, we zoom in on the areas of referential overlap and explore the factors that influence the
degree of lexical heterogeneity and the choices of basic and non-basic color names for a specific
referential area. For instance, as a conceptual factor, is there more or less diversity in one area of the
color spectrum than in the other, and, as an additional contextual factor, is this effect the same
across the four product categories?

9
The multivariate (regression) techniques used in the analysis allow us to give an integrated
multifactorial account for the interaction of conceptual and contextual factors in color category
construal and thus take the notion of near-synonymy in color-naming beyond referential overlap.

Bibliography

Langacker, Ronald W. 2009: A dynamic view of usage and language acquisition. Cognitive
Linguistics 20(3): 627-640.
Langacker, Ronald W. 2009: Investigations in Cognitive Grammar. Berlin, New York: Mouton de
Gruyter.
Divjak, D. 2006. Delineating and Structuring Near-Synonyms. Corpora in cognitive linguistics, St.
Th. Gries & A. Stefanowitsch (eds), Berlin: Mouton, 19-56.
Arppe, A. 2008. Univariate, bivariate and multivariate methods in corpus-based lexicography - a
study of synonymy. Dept. of General Linguistics, University of Helsinki,
http://urn.fi/URN:ISBN:978-952-10-5175-3.

10
Arppe Antti and Divjak Dagmar
University of Helsinki & University of Alberta
University of Sheffield

Synonymy is both Gradient and Context-Dependent

Regardless of the semantic theory or model, in practically all treatments of synonymy the
degree of semantic similarity among words is represented as a constant property – be it Boolean or
gradient – holding among a selected set of words (often pairs), throughout an aggregate of all
conceivable usage contexts. Consequently, little attention is paid to the degree of synonymy,
semantic similarity, or interchangeability among purported synonyms in the underlying multitude of
individual distinct contexts of usage. We argue that in addition to synonymy being a gradient
phenomenon (cf. Miller & Charles 1991), synonymy can be seen to vary in natural usage from one
context to another, such contexts being constituted by explicitly identifiable linguistic properties
that can be analyzed according to some linguistic theory. This view is in line with a probabilistic
theory of how language works.
Our view results from the computational analysis of four Finnish THINK verbs (3,404
sentences extracted from corpora; Arppe 2008) and six Russian TRY verbs (1,351 sentences;
Divjak 2010). We explicitly studied the impact of a multitude of linguistic and extra-linguistic
properties on the selection of one synonym over others in a large number of contexts. In our
analysis, we applied polytomous logistic regression, which models the proportions of occurrence
among a number of possible outcomes, given the occurrence of explanatory properties in various
observable contexts (Arppe 2008). This statistical modeling approach implies that, given a linguistic
analysis scheme, the distributions of expected probabilities of occurrence among possible
alternative outcomes, i.e. synonyms, may in principle vary from one combination of explanatory
properties, i.e. a context, to another. In practice, this variation ranges from a near-categorical
preference for only one of the words in the synonym set, to near-equiprobable contexts where all
synonyms are in principle equally likely to occur, with a variety of other recognizable scenarios of
expected occurrence in between these extremes. Thus, in terms of the explanatory properties, in
practice, some contexts do not allow for synonymy at all, whereas others do not provide much
explicit distinction among the synonymous alternatives. In the latter case, more information about
e.g. the general discourse situation and its participants, including “nuances” outside traditional
linguistic analysis (Inkpen & Hirst 2006), would appear necessary.
Thus, synonymy clearly appears to be both a context-dependant and a contextually varying
phenomenon (cf. Sinclair’s [1991] notion of variation of contextual preferences among word-forms
of the same lemma). It does, nevertheless, vary systematically in terms of the context, and we will
provide concrete example cases of this for both Finnish and Russian.

Bibliography

Arppe, Antti. 2008. Univariate, bivariate and multivariate methods in corpus-based lexicography –
a study of synonymy. Publications of the Department of General Linguistics, University
of Helsinki, No. 44.
Divjak, Dagmar. 2010 (in press). Structuring the Lexicon: a Clustered Model for Near-Synonymy.
Mouton de Gruyter: Berlin/New York.
Inkpen, Diana & Graeme Hirst 2006. Building and Using a Lexical Knowledge-Base of Near-
Synonym Differences. Computational Linguistics 32:2, 223-262.
George A. Miller & Walter G. Charles. 1991. Contextual correlates of semantic similarity.
Language and Cognitive Processes, 1-28.

11
Sinclair, John 1991. Corpus, Concordance, Collocation. Oxford : Oxford University Press.

12
Backus Ad and Mos Maria
Tilburg University
Tilburg University

Near synonymy in partially specific constructions as evidence for the lexicon-


syntax continuum

Rethinking the lexicon-syntax dichotomy as a continuum, as proposed in Construction


Grammar and associated theories, raises the question to what extent there may be synonymy across
what used to be seen as distinct modules. In this contribution, we discuss two Dutch constructions
that express a very similar meaning, but are outwardly very different, one being more
‘morphological’ and the other more ‘syntactic’. The main questions we address is whether the same
factors play a role in lexical and syntactic (near-) synonymy, particularly in the distribution of these
two partially specific constructions, and to what extent differences in distribution reflect differences
in constructional meaning.
Examples of the V-BAAR and IS TE V constructions are given in (1) and (2) respectively. In
both cases, the construction expresses the (lack of) possibility that an entity X (expressed as the
subject) undergoes event Y (expressed as the verbal stem in (1) and in the infinitival form in (2).

1. Zijn handschrift is niet lees-baar


His handwriting is not read-able

2. Zijn handschrift is niet te lezen


His handwriting is not to read.INF

A corpus analysis of instantiations of the constructions shows that both are rather frequent (>
200 types for each construction in the 10-million word Corpus of Spoken Dutch). Out of these, 55
verbs occur with both constructions; instantiations with these verbs therefore potentially constitute
pairs of synonyms. The corpus data and an experiment using novel instantiations showed that the
main meaning difference lies in the speaker’s stance: the IS TE V construction tends to express the
assessment of possibility, whereas the V-BAAR construction is often used for more factual
statements.
Of the 55 verbs, a smaller subset shows signs of lexicalization, in one or both of the
constructions. In such cases, the form has a specific meaning that cannot easily be deduced
compositionally. For example, the verb eten (to eat) produces the forms eetbaar (‘edible, not
poisonous’) and is te eten ‘is tasty’. Usually, the differences are less dramatic: the verb is the
bereiken (can be reached) can refer to both physical and mental destinations, but bereikbaar
(reachable) is nearly exclusively found with concrete locations.
The immediate syntactic context also exerts influence: the IS TE V construction is more
commonly found with verbs that prefer clausal objects. The effect is that pragmatic associations
differ for the two constructions, and to the extent that pragmatics is seen as part of meaning, this
reduces the synonymy between the two. However, at least one factor that plays a role, priming, has
nothing to do with meaning: the sheer occurrence of one of the constructions makes its
reappearance later in the same conversation more likely.
We conclude two things. First, novel instantiations and some of the distributional data provide
evidence that the two constructions show subtle semantic differences. Second, similar factors apply
to both the morphological and the syntactic construction, which argues against maintaining a strict
dichotomy between lexicon and syntax.

13
14
Borin Lars and Forsberg Markus
Språkbanken, Dept. of Swedish Language
Språkbanken, Dept. of Swedish Language

Beyond the synset: Swesaurus -- a fuzzy Swedish wordnet

Swesaurus is a free Swedish wordnet currently under construction in our research unit.
Swesaurus is made up in part by a combination of a number of pre-existing freely available lexical
resources. Two central resources for the purposes of this presentation are SALDO and Synlex.
SALDO (Borin et al. 2008; Borin & Forsberg 2009; <http://spraakbanken.gu.se/eng/saldo/>)
is a full-scale Swedish lexical-semantic resource with non-classical, associative relations among
word and multiword senses. The senses in SALDO are identified by carefully designed persistent
formal identifiers, and for this reason, SALDO has become the “pivot” resource of all our
computational lexicon activities, including Swesaurus (Borin et al. 2010;
<http://spraakbanken.gu.se/eng/swefn/>). We will say something about the relationship of
SALDO’s associative relations to classical lexical-semantic relations and how we intend to
incorporate the SALDO relations in Swesaurus.
Synlex (Kann & Rosell 2006) is a graded Swedish synonym list created by asking members
of the public – users of an online Swedish-English dictionary – to judge the degree of synonymy of
a random, automatically generated synonym pair candidate.
We will describe our experiments turning the graded synonymy relations of Synlex into fuzzy
synsets in Swesaurus. The introduction of fuzziness into a wordnet raises many intricate
methodological and theoretical questions, e.g., if w1 is a graded synonym of w2, and w1 is a
hyponym of w3, what is the relation between w2 and w3? Similarly, if wa is a synonym of degree
0.75 of wb and wb is a synonym of degree 0.9 of wc, what – if any – is the degree of synomymy
between wa and wc?
If synonymy is seen as an all-or-none transitive and symmetric relation, the construction of
synsets from synonym pairs is arguably straightforward: We can simply compute the transitive
closure of the synonymy relation.
When graded synonymy enters the picture, which method to use for collecting synonym pairs
into synsets becomes much less obvious, and especially how to assign a degree of synonymy to
“derived” pairs, i.e., pairs not in the original list. We have experimented with two different
approaches for turning Synlex synonym pairs into fuzzy synsets in Swesaurus, transitive closure
and clique formation. We will present the outcomes of these experiments and discuss the merits and
disadvantages of the two methods.

Bibliography

Borin, Lars and Markus Forsberg 2009. All in the family: A comparison of SALDO and WordNet.
Proceedings of the Nodalida 2009 Workshop on WordNets and other Lexical Semantic
Resources – between Lexical Semantics, Lexicography, Terminology and Formal
Ontologies. Odense: NEALT. 7–12.
Borin, Lars, Markus Forsberg and Lennart Lönngren 2007. The hunting of the BLARK – SALDO,
a freely available lexical database for Swedish language technology. Resourceful
language technology. Festschrift in honor of Anna Sågvall Hein, Joakim Nivre, Mats
Dahllöf and Beáta Megyesi (eds). Acta Universitatis Upsaliensis: Studia Linguistica
Upsaliensia 7. 21–32.
Borin, Lars, Dana Danélls, Markus Forsberg, Dimitrios Kokkinakis, Maria Toporowska Gronostaj
2010. The past meets the present in Swedish FrameNet++. Proc. of Euralex 2010.

15
Kann, Viggo and Magnus Rosell 2006. Free construction of a free Swedish dictionary of synonyms.
Proc. of the 15th NODALIDA. Dept. of Linguistics, University of Joensuu. 105–110.

16
Calude Andreea and Pagel Mark
University of Reading
University of Reading

Selfish Words: what can words in competition tell us about language change?

As speakers, we have a number of word choices at our disposal when it comes to expressing a
given meaning, for example, in English, when thinking of the concept of “two”, one can refer to this
by saying “pair”, “twin”, “two”, “couple” and so on, yet, most often we chose “two” over all other
lexical items. What governs the choices we make and how can we characterise this linguistic
behaviour? Using techniques devised for describing population genetics and evolutionary biology,
our results show that like genes, words come under pressure (that is, biased in some way), rather
than being used neutrally under drift.
In the paper, we explore lexical variation in two corpora of Spoken American English
(namely from LAMSAS, Kretzschmar et al 1993, and LAGS, Pederson et al 1986) using statistical
tools to capture word-use within populations of speakers. The data consist of spoken records
characterising the use of words for everyday items, concepts or events such as “wife”, “chair”,
“thunderstorm”, “driven”, or “three”. The participants were asked identical sets of questions like
“what do you call the place where you store your clothes?” thereby aiming to elicit responses such
as “closet”, “wardrobe” and other variants. Here we report on the frequencies of different words
which participants select to refer to the same items, for example, “attic”, “loft”, “garret”, and so on.
Our interest, located within a cognitive approach to language and cultural evolution is in whether
the choice of word is proportional to how often each word is used among the population of speakers
(termed here “neutral use”), or whether word use is biased in favour of the most frequently used
words (“conformist use”).
Using Ewens Sampling Formula (1972), we find that the majority of these word-use
frequency distributions deviate significantly from what would be expected of “neutral use”. In all of
these cases one or a small number of variants (e.g., “attic”) dominates the distribution, being used
far more often than expected by chance given the availability of other forms (such as “garret” or
“loft”). In the remaining cases, the different word-variants (for instance, “fire-grates” or “andirons”)
are used in frequencies expected under our neutral-use scenario and in no case were different words
employed in equal or even approximately equal frequencies.
The current work explores the concept of "synonymy" from a usage-driven perspective, where
words which occupy (at least) some overlapping semantic space (variants such as "attic" and "loft")
can be understood to compete against the each for the attention of speakers in a given population
(much like viruses or other biological phenomena, Mufwene 2002). We show how this approach
can sharpen our understanding of not just how language-use evolves within communities but why
some variants come to dominate.

Bibliography

Ewens, W. (1972). The sampling theory of selectively neutral alleles. Theoretical Population
Biology 3: 87–112.
Kretzschmar, W., McDavid, V., Lerud, T.K. and Johnson, E. (ed.) (1993). Handbook of the
Linguistic Atlas of the Middle and South Atlantic States. Chicago: The University of
Chicago Press.
Mufwene, S. (2008). Language Evolution: Contact, Competition and Change. Continuum.
Pederson, Lee, Susan Leas, Guy Bailey, and Marvin Bassett. (1986). The Linguistic Atlas of the
Gulf States. Vol. 1, Handbook. Athens: Univ. of Georgia Press.

17
Cappelle Bert and Desutter Gert
University College Ghent and Ghent University
University College Ghent and Ghent University

We ought to re-think synonymy, shouldn’t we?


A corpus-based reassessment of two English modals’ substitutability

The English modals should and ought to “seem to be largely interchangeable” (Palmer 1990; cf.
similar sentiments in Coates 1983: 69; Quirk et al. 1985: 227; Huddleston and Pullum 2002: 186).
Some proposals for possible distinctions have nonetheless been made, based inter alia on (i)
subjectivity vs. objectivity (Swan 1980: 550; Gailor 1983: 348; Aarts and Wekker 1987: 193;
Myhill 1996; Collins 2009: 54) (ii) absence vs. presence of an implication of non-fulfillment (Close
1981: 121; Gailor 1983: 348-349; Westney 1996: 170) and (iii) relative frequency vs. infrequency
of epistemic reading (Coates 1983; Palmer 1987: 134; Collins 2009).
To test whether these and a few other claims made in the literature could be verified, we
conducted a corpus-based study involving over 1000 sentences with should or ought to in
contemporary spoken and written British English and coded them for 28 parameters. Raw frequency
and frequency per discourse mode could not be coded, but the data confirmed previous observations
that should is used much more frequently than ought to and that the frequency of ought to in spoken
language is higher than its frequency in written language. To find out which of the parameters
investigated contribute significantly to the distribution of should and ought to, we fitted a logistic
regression model, making sure any corruption of the model caused by multicollinearity was
excluded.
Eight parameters were shown to exert a unique and significant impact on the choice of should vs.
ought to, in order of decreasing strength: (i) inversion, (ii) a following contracted perfect infinitive,
(iii) no adverb or a following rather than preceding adverb, (iv) verb-marked negation, (v)
embedding by suggest or a similar item, (vi) reference to the non-past, (vii) no embedding by think
or a similar cognition expression and (viii) third (vs. first or second) person subject. All other
factors turned out not to be significant or were left out of the model because they correlated with
one or more parameters in the model.
Our model thus provides a parsimonious and accurate description of the should/ought to
variation. It also guides us towards a possible explanation, which we believe can be found in the
different grammatical status of should and ought to, the former being a true modal auxiliary, the
latter being a semi-auxiliary (a blend between a modal and a lexical verb). We argue that this
difference can account for the four strongest distinctions in usage we revealed.
We further discuss the results of a subsequent distinctive collexeme analysis (Stefanowitsch and
Gries 2003) supporting the claim made above that should (compared to ought to) occurs
relatively more frequently with situations which are likely to be carried out than with unlikely
situations (cp., e.g., It should be noted that… vs. He ought to be crucified) but rejecting the above
claim that should is more subjective (e.g. Abstracts should / ??ought to be submitted by April 16,
2010). Finally, we reflect on the implications of our study for the (non)existence of free variation in
language.

18
Bibliography

Aarts, Flor and Herman Wekker. 1987. A Contrastive Grammar of English and Dutch –
Contrastieve Grammatica Engels/Nederlands. Leiden: Martinus Nijhoff.
Close, R. A. (1983) English as a Foreign Language: Its Constant Grammatical Problems, 3rd ed.,
London: George Allen & Unwin.
Coates, Jennifer (1983) The Semantics of the Modal Auxiliaries. London and Canberra: Croom
Helm.
Collins, Peter (2009) Modals and Quasi-Modals in English. Amsterdam and New York: Rodopi.
Gailor, Denis (1983) “Reflections on Should, Ought to, and Must,” English Language Teaching
Journal 37: 346-349.
Huddleston, Rodney and Geoffrey K. Pullum (2002) The Cambridge Grammar of the English
Language. Cambridge: Cambridge University Press.
Myhill, John (1996) “Should and Ought: The Rise of Individually Oriented Modality in American
English.” English Language and Linguistics 1: 3-23.
Palmer, Frank Robert (1987) The English Verb, 2nd ed. London and New York: Longman.
Palmer, Frank Robert (1990) Modality and the English Modals, 2nd ed. London and New York:
Longman.
Quirk, Randolph, Sidney Greenbaum, Geoffrey Leech and Jan Svartvik (1985), A Comprehensive
Grammar of the English language. London: Longman.
Stefanowitsch, Anatol and Stefan Th. Gries. 2003. Collostructions: Investigating the interaction
between words and constructions. International Journal of Corpus Linguistics 8: 209-
243.
Swan, Michael (1980) Practical English Usage. Oxford: Oxford University Press.
Westney Paul (1995) Modals and Periphrastics in English: An Investigation into the Semantic
Correspondence between Certain English Modal Verbs and Their Periphrastic
Equivalents. Tübingen: Niemeyer.

19
Copestake Ann and Herbelot Aurélic
Computer Laboratory, University of Cambridge
Computer Laboratory, University of Cambridge

Synonymy in an approach to combined distributional and compositional


semantics

We outline a perspective on synonymy from a novel approach to combining distributional and


compositional semantics, Lexicalized Compositionality (LC). In LC logical forms, the standard
model-theoretic lexical predicates (e.g., catʹ) are replaced with distributional concepts (notated as
e.g., cat°). cat° corresponds to the set of all linguistic contexts (expressed as logical forms) in
which the lexeme cat occurs. We consider hypothetical contexts, which are based on all the
possible statements true in some world (allowing a comparison to model-theoretic approaches), and
actual contexts, corresponding to an individual's linguistic experience.
The aim of LC is to investigate models of semantics which are psycholinguistically plausible
in terms of linguistic information. Some utterances are directly grounded. In such cases, contexts
can be paired with perceptual data, and linguistic concepts, such as cat°, related directly to real
world entities. However utterances can be understood which are not immediately grounded at all or
only partially grounded. We assume that speakers' knowledge of word meaning is primarily
acquired implicitly, from actual contexts (grounded or ungrounded), but that there is also a more
minor role for explicit `expert' knowledge. Individual speakers will have somewhat different
models of word meaning, and understanding may thus require negotiation.
Within this approach, the notion of synonymy is multifaceted. Classic synonymy corresponds
to the situation where a hearer is (more or less) directly informed that a term means the same thing
as one they are already familiar with. For example: The aubergine (eggplant) has to be one of my
favourite vegetables. http://www.earth.li/~kake/cookery/aubergine.html (last accessed 10th August
2010)
The much commoner case of near-synonymy (Edmonds and Hirst, 2002) corresponds to
similarities between context sets. We hypothesise that the process of understanding a new word
involves an initial approximation to the nearest known linguistic concepts(s) (by context set
comparison), followed by the acquisition of differentiating information. We define a notion of a
characteristic context in terms of the differences between two context sets. For instance, a hearer to
whom rancid is a relatively unfamiliar term may initially relate it to off/bad/rotten etc, but then
additionally differentiate it using further contexts (perhaps by its preferential application to fatty
foodstuffs compared to bad). Alternatively, or additionally, an `expert' definition may be known.
In normal communication, the coarse-grained notion of synonymy (i.e., that rancid relates to off etc)
usually suffices for understanding.
We should emphasize the differences between our approach and the usual computational
models of distributional similarity (Harper, 1965, onwards). Clearly our contexts are much more
structured: the recent trend is towards higher structure (e.g., Erk and Padó (2008)), but our work is
still distinct in aiming for a logical account that incorporates distributions, thus allowing for
(relatively conventional) logical inference. Furthermore, in our account, the contexts come from a
specific individual's experience. We expect this to give very different results from the standard
approaches which utilise data from very heterogeneous sources.
Our work is at a very early stage and has primarily involved development of the theoretical
account. Given the current lack of individualised corpora, empirical work involves approximations
of linguistic contexts. However, we believe these ideas suggest new theoretical perspectives for
computational experiments on synonymy.

20
Bibliography

Edmonds, Philip and Graeme Hirst (2002) ‘Near-synonymy and lexical choice’, Computational
Linguistics, 28(2), 105–144.
Erk, Katrin and Sebastian Padó (2008) ‘A Structured Vector Space Model for Word Meaning in
Context’, Proceedings of the EMNLP, Honolulu, HI.
Harper, K.E. (1965) ‘Measurement of similarity between nouns’, The RAND Corporation, Santa
Monica.

21
Dalmas Martine and Dobrovol’skij Dmitrij
Université Paris-Sorbonne
Russian Academy of Sciences

A corpus-based analysis of quasi-synonymous adjectives in German

The aim of this paper is to prove to what extent corpus-based statistical data is helpful for
semantic and pragmatic discrimination of quasi-synonymous adjectives, even if these adjectives are
represented in dictionaries as meaning absolutely the same. The analysis of combinatorial profiles
of a group of quasi-synonymous German adjectives (namely ausgezeichnet, hervorragend,
exzellent, herrlich, vortrefflich and vorzüglich) allows us to objectivate both the semantic structure
of nomina modified by a given adjective and the discourse domain. Since these two parameters
govern the use of every adjective in question such an analysis can provide an appropriate basis for
the relevant semantic and pragmatic discrimination.
In principle, there are three ways to explain relevant specifics of the combinatorial profiles of
these adjectives.
(a) There are combinatorial differences which go back to the semantic class of the modified
nouns. In this case, the semantic structure of the adjective itself displays features that predict
combinatorial restrictions and preferences, i.e. the specifics of co-occurrence are semantic in nature.
(b) There are differences which go back to given discourse domains. In this case, the specific
properties of the combinatorial behaviour of every adjective go back to their preference for certain
discourse types or registers as well as for certain thematic fields, rather than to the meaning of the
adjective itself. So, there are words that are avoided in spoken colloqual language, but are
frequently used in the feuilleton of big magazines. In the everyday language people talk about
weather differently than in the weather report. On the other hand, the choice of an adjective from
the given group of quasi-synonyms can be influenced by the subject area discussed in the text. For
instance, people talk about sports in different terms than about antique arts or about food. It means
that besides the discourse type and register also the thematic domain plays a role while choosing a
fitting word. Often it is not possible to separate both phenomena from each other (a reportage on
sport events is, e.g., a thematically bound discourse type). Therefore, we consider both phenomena
within the same category, namely discourse domain. The meaning explanations of the adjectives in
question do not display any differentiative semantic features. They can be discriminated from each
other only by an additional comment, such as “preferably in discourse domain X”.
c) There are also combinatorial peculiarities which cannot be explained systematically, i.e. the
corpus analysis clearly shows that a given adjective may have combinatorial preferencies that go
back neither to its semantic class nor to a certain discource domain. Here we are dealing with
individual combinatorial properties that have to be fixed as lists. What is extremely important is that
the list-representation makes sense only if all other possibilities do not work.
Which of the three explanation ways (a), (b) or (c) fits the data in every concrete case can be
decided only on the basis of corpus analysis. We used here the platform CCDB Mannheim (© C.
Belica) and the tools of the DWDS-Wortprofile (Berlin-Brandenburg Academy of Sciences).
The first phase of our research (Dobrovol’skij 2004; Dalmas/Dobrovol’skij in print) has
revealed significant combinatorial differences among the given lexemes, which can partly be
explained via properties of their inner form fixed in their morpheme structures. Conceptual features
of this kind have to be included in the semantic explanation.

22
Bibliography

Biber, Douglas/ Conrad, Susan/ Reppen, Randi (1998): Corpus linguistics: investigating language
structure and use. Cambridge: Cambridge University Press.
Dalmas, Martine/ Dobrovol’skij, Dmitrij (in print): Quasisynonymie bei Adjektiven: hervorragend
und Co (eine corpusgestützte Untersuchung). In: Schmale, Günter (Hg.): Das Adjektiv.
Was wird wo, wie und wozu „dazugeworfen“?. Tübingen: Stauffenburg.
Dobrovol’skij, Dmitrij (2004): Corpusbasierte kontrastive Lexikologie: semantische und
kombinatorische Aspekte. In: Mehrsprachige Individuen – vielsprachige
Gesellschaften. Abstracts zur 35. Jahrestagung der Gesellschaft für Angewandte
Linguistik in Wuppertal. Wuppertal: Bergische Universität Wuppertal. 33-34.
Hanks, Patrick (2008): Lexical patterns: from Hornby to Huston and beyond. In: Proceedings of the
XIII EURALEX International Congress. Barcelona: IULA, 2008. 89-129.
Jones, Karen Sparck (1986): Synonymy and semantic classification. Edinburgh: Edinburgh
University Press.
НОСС – Новый объяснительный словарь синонимов русского языка. 2-е изд., испр. и доп. /
Под рук. Ю.Д. Апресяна. Москва/Вена, 2004.

23
Deshors Sandra C.
University of Sussex

Semantic similarity in non-native English:


the case of may and can in French-English interlanguage

This work presents one possible empirical route to trace patterns of semantic behavior in a
corpus of French learner English. I investigate to what extent the expression of the semantic domain
of POSSIBILITY, as conveyed by may and can, differs in native English and French-English
interlanguage. I will specifically discuss (i) how English native speakers use may and can in a more
data-driven way than is usually adopted in studies on modality, and (ii) how the uses of can and
may by French learners differs from that of native speakers and what these distributional differences
suggest as to what motivates the different patterns.
At the intersection between corpus linguistics and cognitive semantics, Gries & Divjak (2009)
and Gries & Otani (2010) argue in favor of a Behavior Profile (BP) approach to semantic analysis,
which allows to explore how meanings and functions of lexical and syntactic elements are
correlated with the distribution(s) of formal elements within their contexts. However, with only one
exception, applications have focused on polysemy, synonymy, and antonymy to native-speaker data
from one language only or, in the case of Divjak & Gries (2009), the study of phasal verbs in
English and Russian.
The current work presents an application of the BP approach to the domain of interlanguage. I
investigated 3700 instances of may and can in English and French-English interlanguage as well as
French pouvoir in three different corpora: the French sub-section of the International Corpus of
Learner English (ICLE), the Louvain Corpus of Native English Essays (LOCNESS) and the Corpus
de Dissertations Françaises (CODIF). I annotated these instances for 22 semantic and morpho-
syntactic variables, including a total of 98 features and resulting on tens of thousands of data points.
The statistical treatment of the annotated data includes both a monofactorial analysis that involves
assessing the behavior of may and can in relation to individual semantic and morpho-syntactic
independent variables; and a multifactorial analysis which involves the statistical modelling of
possible interactions between independent variables and their effects on may and can.
The results of the monofactorial assessment confirm the relevance of a contextually-grounded
approach. Linguistic components such as the senses of may and can, the type of clauses they occur
in, the type of lexical verbs they occur with or the type of (in)animacy of their subject referents
have all been identified as highly influencing components within the corpus (each individual
component showing a p-value smaller than 0.001). However, results also indicate that the
frequencies of may and can differ considerably across the corpora (chi-square= 3716.93, p<0.001,
Cramer's V=0.71), thus suggesting that the identified semantic and morpho-syntactic features yield
different interaction patterns in each individual sub-corpus. I am currently carrying out a
multifactorial analysis to investigate behavioral patterns characteristic of may and can both in L2
and IL.
This work shows that, in addition to providing a data-driven methodology for linguistic
patterns and their semantics in one language, the BP approach also provides a useful empirically-
grounded way to study characteristics of interlanguage varieties.

Bibliography

Gries, Stefan Th. and Dagmar S. Divjak. 2009. Behavioral profiles: a corpus-based approach to
cognitive semantic analysis. In New directions in cognitive linguistics, ed. by Vyvyan
Evans and Stephanie S. Pourcel, 57-75. Amsterdam, Philadelphia: John Benjamins.

24
Gries, Stefan Th. & Naoki Otani. 2010. Behavioral profiles: a corpus-based perspective on
synonymy and antonymy. ICAME Journal.

25
Devos Maud
Royal Museum for Central Africa

Two times ‘go’ in Shangaci: semantics or pragmatics?

Shangaci, a Bantu language spoken in Mozambique, has two seemingly synonymous verbs to
express ‘go’: -entta and -lawa. In the question below the two ‘go’-verbs are perfectly
interchangeable.

(1) a. o-tt-éntt’ ó-nliímpu


2SG-PRES-go 17-3.well
‘are you going to the well?’

b. o-ttí-láw ó-nliímpu
2SG-PRES-go 17-3.well
‘are you going to the well?’

However, the answers to the above questions suggest that the verbs differ in meaning and/or
use.

(2) a. waalá ki-tt-éntt’ ó-muúti


nor 1SG -PRES -go 17-3.town
‘no, I am going home’

b. kha-n-cúuwi nkhámá ki-ttí-laáwa


NEG-1PL-know if 1SG-PRES-go
‘I do not know whether I am going’

Moreover, -entta and -lawa follow very different grammaticalization paths. The verb -entta is
used as an auxiliary to express predication focus (3) (Devos and van der Wal 2010), whereas the
auxiliary -lawa is used for discourse deixis (4) (Bourdin 2008) indicating that an action takes place
at some distance from the deictic centre. As persistence of the original lexical meaning is typical in
grammaticalization (Hopper 1991), the different paths followed by -entta and -lawa suggest a
difference in lexical meaning.

(3) yoómbw’etó kha-y-iital-éeni y-éntt-é ó-khízeéy-a


7.pot 7.DEM NEG-7-be.full- PERF 7-GO-PERF 15-be.halffull-INF
‘that pot is not full, it is half full'

(4) a-tthíir-i m-muuti’ phúule a-láa a-cí-víith-i


2-run-PERF 18-3.town 18.DEM 2-GO 2-REF-hide-PERF
‘They fled from that town, and hid themselves’

The paper aims at defining the nature of the difference between the two verbs (pragmatic or
semantic) and its implication for their identification as synonymous or merely similar forms. A
related theoretical question is whether lexemes can be considered synonymous when they are
semantically identical but pragmatically different, or reformulated: should synonyms be defined in
purely semantic terms?

26
Botne (2005) identifies a similar pair of ‘go’-verbs in the Bantu language Chindali. According
to him the difference is of a purely pragmatic nature: both verbs express going somewhere but in
one case the goal is salient, whereas in the other case the motion is salient. The Shangaci verbs
could be said to involve the same pragmatic difference. In (2a) the goal is contrastively focused
whereas in (2b) the going itself is questioned. Moreover, the grammaticalization of -entta as a
marker of predication focus involves strengthening of the salience of goal. If the difference is solely
pragmatic in nature, -entta and -lawa could be said to be synonymous forms sharing the same
image-schematic structure.
However, the difference in use of -entta and -lawa could also be a consequence of a more
basic semantic distinction. As argued by Wilkins and Hill (1995) not all ‘go’- verbs are deictic and
Shangaci could be said to have a deictic (i.e., -lawa) as well as a non-deictic (i.e., -entta) ‘go’-verb.
In the grammaticalization of -lawa as a marker of discourse deixis, it is the deictic component of its
meaning (away from deictic centre) that is strengthened. Such a semantic difference would imply
that -entta and -lawa are similar but not synonymous forms.

Abbreviations

DEM demonstrative
INF infinitive
NEG negative
PERF perfective
PL plural
PRES present
REF reflexive
SG singular

Bibliography

Botne, Robert. 2005. “Cognitive schemas and motion verbs: COMING and GOING in Chindali
(Eastern Bantu)”. Cognitive Linguistics 16(1).43–80.
Bourdin, Philippe. 2008. “On the grammaticalization of ‘come’ and ‘go’ into markers of textual
connectivity’. In Rethinking Grammaticalization: New Perspectives, López-Couso,
Maria José and Elena Seoane (eds.). 37-59. Amsterdam/Philadelphia: John Benjamins.
Devos, Maud and Jenneke van der Wal. “‘Go’ on a rare grammaticalization path to focus”. TiNdag
2010, Utrecht.
Hopper, Paul J. 1991. “On some principles of Grammaticalization”. In Approaches to
Grammaticalization, vol. 1. Traugott, Elizabeth Closs and Bernd Heine (eds.). 17-35.
Amsterdam: Benjamins.
Wilkins, David P. and Deborah Hill. 1995. “When “go” means “come”: Questioning the basicness
of basic motion verbs”. Cognitive Linguistics 6(2/3). 209-259.

27
Faulhaber Susen
Friedrich-Alexander-University Erlangen-Nuremberg

To what extent is synonymy a good indicator for verb complementation?


Challenging semantics-based accounts

This paper will discuss the relationship between verb meaning and verb complementation
on the basis of a corpus-based comparison of semantically similar verbs (Faulhaber
forthcoming). The question addressed is whether syntactic information can be deduced in a regular
way from the meaning of verbs or has to be regarded as a matter of storage.
It is undoubtedly the case that meaning plays a role – and most likely even an important one –
in how activities or situations are verbalized, i.e. with which formal complement types a verb can be
combined or, alternatively, in which valency patterns it can occur. This is reflected in statements
such as that “[k]nowing the meaning of a verb can be a key to knowing its behavior” (Levin 1993:
5), that there is “a principled interaction between the meaning of a word and its grammatical
properties” (Dixon 1991: 6) and that “[s]emantically similar verbs show a strong tendency to appear
in the same argument structure constructions” (Goldberg 2006: 58), which can be found in the
context of various linguistic theoretical frameworks. Evidence provided for this assumption ranges
from hand-picked examples to broad comparisons of verbs and their complementation patterns such
as Levin’s English Verb Classes and Alternations (1993) or Hunston and Francis’s Pattern
Grammar (2000), which aim to show that verbs which exhibit similar grammatical behavior – in the
sense that they allow the same diathesis alternations (Levin 1993) or can occur in the same
complementation patterns (Hunston and Francis 2000) – are also semantically similar. By
implication, it could be (and has been) construed that synonymous or semantically similar verbs are
also syntactically alike.
However, this assumption is clearly not without problems. For once, the conclusion that
semantic similarity leads to syntactic similarity cannot be based on studies which take syntactic
similarity as a starting point since differing uses are excluded ex ante (cf. Levin 1993 or Hunston
and Francis 2000). Moreover, counter-evidence is not as “rare” as is often assumed, which becomes
evident when large amounts of actual language data are taken into consideration. A comparison of
more than twenty groups of semantically similar English verbs (all in all more than eighty verbs) as
regards their complementation patterns – based on the corpus-based Valency Dictionary of English
(Herbst, Heath, Roe, and Götz 2004) and further corpus research – clearly reveals that despite clear
structural similarities a semantic determinist view of verb complementation is highly problematic:
The amount of idiosyncratic restrictions, i.e. restrictions which can neither be accounted for on the
basis of verb meaning (e.g. the verbs’ participants, lexical aspect, selection restrictions etc.) nor of
the meaning of the complement types or patterns, is too big to be dismissed as peripheral (Faulhaber
forthcoming). These results clearly emphasize the role of conventionalization in verb
complementation, providing a strong case for rethinking a deterministic role of semantics in verb
complementation and for strengthening the role of storage instead.

Bibliography

Dixon, Robert M.W. 1991. A New Approach to English Grammar, on Semantic Principles. Oxford:
Clarendon Press.
Faulhaber, Susen. forthcoming. Verb valency patterns – Challenging semantics–based accounts.
Berlin/New York: Mouton de Gruyter.
Goldberg, Adele. 2006. Constructions at Work. The Nature of Generalization in Language. New
York: Oxford University Press.

28
Herbst, Thomas, David Heath, Ian Roe and Dieter Götz. 2004. A Valency Dictionary of English.
Berlin/New York: Mouton de Gruyter.
Hunston, Susan and Gill Francis. 2000. Pattern Grammar. A corpus-driven approach to the lexical
grammar of English. Amsterdam: Benjamins.
Levin, Beth. 1993. English Verb Classes and Alternations – A Preliminary Investigation. Chicago:
The University of Chicago Press.

29
Gaillard Benoît, Navarro Emmanuel and Gaume Bruno
University of Toulouse 2
University of Toulouse 2
University of Toulouse 3

From binary synonymy to near synonymy by optimal proxemy of lexical


resources

Dictionaries of synonyms are often encoded as binary links between words. This encoding
raises many issues. For example, we show that there is very little agreement between different
expert-built resources, whereas they represent a similar linguistic reality. This may be due to the
fact that absolute synonyms are rare and that most of the synonymy described in the dictionaries is
in fact a description of nuances of a general meaning shared by a set of near-synonyms (Edmonds
and Hirst, 2002). The complexity and scale of these nuances leave room for interpretation when
projecting them on the basic “synonym/not synonym” alternative, which in turn gives rise to
discrepancies between electronic lexical resources at the most fine-grained level. However, we
claim that, at a coarser level, the patterns drawn by sets of near-synonyms should be mostly
independent from the resource used to describe the lexicon of a given language. We propose a
model that leverages the binary encoding of resources to represent the synonymy structure at
various levels of granularity. This is not a new merging method, but a attempt at a better modelling
of the notion of synonymy. Based on this model, we find an optimum level of granularity for which
resources are the most similar, therefore our approach seems appropriate in order to infer near-
synonymy patterns from any particular binary encoded resource.
We studied the similarities of seven well-known french dictionnaries, binarily encoded as
graphs in which vertices are words and two vertices are linked by an edge if they are considered
synonyms. We used the F-score between sets of edges to measure the similarities between the
unweighted graphencoded dictionaries of synonyms and found that it does not exceed 0.5. To model
these resources on coarser levels, we introduce a notion of proxemy between any two words, based
on the probability of reaching one word from another after a random exploration of the graph. At
each step of the exploration, a particle is modelled to move from one word to any of its neighbours
with a probability inversely proportional to its degree. Each word is thus associated to a proxemy
vector characterising its semantic similarity with the other words. This vector depends on the
starting vertex and on the number of exploration steps. When the number of steps tends to infinity,
all words have the same proxemy vector, which describes the importance of each word in the graph,
at the highest level of generality (Gaume et al., 2010). Conversely, after only one step, proxemy
vectors have the lowest level of generality, taking only the neighbours of each node into account.
The distance between two resources is then defined as the mean euclidian distance between the
proxemy vectors of the same word in the two graphs. We experimentally found an optimal number
of proxemy steps for which the similarity of any two resources is maximal. This is an optimal level
of generality, according to the proxemy model, that enables us to discover near-synonymy patterns
that are similar across dictionaries whose binary synonymy patterns are very different.

Bibliography

P. Edmonds and G. Hirst (2002) Near-synonymy and lexical choice, Computational Linguistics,
28(2), pp. 105-144.
B. Gaume, F. Mathieu and E. Navarro (2010) Building Real-World Complex Networks by
Wandering on Random Graphs, Information - Interaction – Intelligence, (To appear).

30
Glynn Dylan
University of Lund

Usage-Based Cognitive Models.


Synonymy as an operationalisation of conceptual structure

Synonymy could be the key to operationalising the study of culturally determined concepts.
As such, this paper aims at developing a method for the corpus-driven analysis of cognitive models.
It focuses on the concept of LIBERTY in 20C Amer-ica. The description of linguistically encoded
culturally rich concepts is a basic enterprise of Cognitive Linguistics (Wierzbicka 1985, Lakoff
1987, 1995). Despite the insights of such research, its method of analysis is notoriously weak -
concep-tual structures are proposed, based on sets of contextless expressions. Firstly, this method
offers no means for the verification of results. Secondly, it offers no infor-mation about the how the
expressions, and therefore the concepts, are used. The actual structure of the concept, relative to a
speech community, is entirely un-known. These “Idealised” Cognitive Models are devoid of the
complexity of social reality. A usage-based approach to concepts must resolve these two
shortcomings. Drawing on Divjak (2006), Janda & Solovyev 2009, and Glynn’s (2010) approach to
lexical synonymy, the study operationalises the problem of conceptual descrip-tion by examining
the use of the lexemes associated with a given concept.
The analysis is based upon two lexemes, examples of which are extracted from Time
magazine - 250 occurrences of liberty and 250 of freedom (25 randomly selected occurrences from
the first three years of each decade). Time magazine is chosen because it represents mainstream
readership in American culture. The dia-chronic dimension of the study will help capture the social
variation that is inherent in a Usage-Based Cognitive Model. Adopting the type of feature analysis
developed in Gries (2006) and Glynn (2009), the examples are annotated for a range of formal,
semantic, and sociolinguistic phenomena. Important factors include the grammatical construction
associated with the lexeme, its source domain in figurative uses, the kind of patient affected by
LIBERTY, its agentivity, as well as the topic of discourse.
Multiple Correspondence Analysis helps identify how the conceptual structure varies. During
the 20s, 30, and 40s, liberty is often used when discussing nations, this shifts during the 50s and 60s
to discussions about society, and then this shifts again during the 70s - 2000s to discussions of
individual liberty. The metaphors associated with the lexeme shift in a similar pattern beginning
with ‘liberty’ being conceptualised as a simple ‘object’ then toward the end of the century being
con-ceptualised as a ‘struggle’ or as a ‘place’. Similar patterns are observed for the use of the
lexeme freedom. An analysis of variance demonstrates that the most signifi-cant and important
variation over the 20C is the kind of experiencer to which the lexemes refer. This runs parallel to
the results of the Correspondence Analysis. A Regression Analysis focuses on the importance of the
experiencer in the conceptu-alisation of LIBERTY and confirms that there is a statistically
significant shift from abstract non-human experiencers of LIBERTY to concrete and individual
experi-encers. Could this represent a shift towards a more individualist conceptualisation or is it a
merely a result of world politics and the political emphasis of the maga-zine? A comparative study,
based on a different genre needs to be performed to answer this question. The study demonstrates
the importance of the method for the study of cognitive models, both because it permits result
verification and because it adds a usage-based dimension to those results.

Bibliography

Divjak, D. 2006. Delineating and Structuring Near-Synonyms. Corpora in cognitive linguistics, St.
Th. Gries & A. Stefanowitsch (eds), 19-56. Berlin: Mouton.

31
Divjak, D. 2010. Structuring the Lexicon: a Clustered Model for Near-Synonymy. Berlin: Mouton.
Glynn, D. & Fischer, K. (eds). 2010. Corpus-Driven Cognitive Semantics. Usage-Based
approaches to conceptual structure. Berlin: Mouton.
Glynn, D. & Robinson, J. (eds). In press. Polysemy and Synonymy. Corpus methods and
applications in Cognitive Linguistics. Amsterdam: Benjamins.
Glynn, D. 2010. Synonymy, Fields, and Frames. Developing usage-based methodology for
Cognitive Semantics. Cognitive Foundations of Linguistic Usage Patterns, H.J. Schmid
& S. Handl (eds)., 89-118. Berlin: Mouton
Gries, St. & Stefanowitsch, A. (eds). 2006. Corpora in Cognitive Linguistics. Corpus-based
Approaches to Syntax and Lexis. Berlin: Mouton.
Gries, St. Th. 2006. Corpus-based methods and cognitive semantics: the many meanings of to run.
Corpora in Cognitive Linguistics. St. Gries & A. Ste-fanowitsch (eds), 57-99. Berlin:
Mouton.
Janda, L. & Solovyev, V. 2009. What Constructional Profiles Reveal About Synonymy: A Case
Study of Russian Words for SADNESS and HAPPINESS. Cognitive Linguistics 20:
367-393.
Lakoff, G. 1987. Women, Fire, and Dangerous Things. What categories reveal about the mind.
London: UCP.
Lakoff, G. 1995. Moral Politics: What Conservatives Know That Liberal Don’t. London: UCP.
Stefanowitsch, A. & Gries, St. (eds). 2006 Corpus-based Approaches to Metaphor and Metonymy.
Berlin: Mouton.
Wierzbicka, A. 1985. Lexicography and Conceptual Analysis. Ann Arbor: Karoma.

32
Glynn Dylan and Levshina Natalia
University of Lund
University of Leuven

The Synonymy of Morphological Semantics.


A usage-based study of long and short adjectives in Russian

Synonymy is not unique to lexical semantics. From a Cognitive Linguistics perspective,


where there is no theoretical distinction between lexis and syntax and all form is semantically
motivated, syntactic and morphological alternations are, in fact, the study of (near) synonymy. The
corpus-driven study of both synonymous syntax and lexis has a strong tradition in Cognitive
Linguistics (Divjak 2006, Arppe 2008, Janda & Solovyev 2009, Glynn 2010, Heylen, 2005,
Grondelaers et al. 2007, Divjak 2009, Speelman & al. 2010, Glynn & Robinson forthc.). However,
Morphological semantics present certain hurdles that are not present in lexical and syntactic
research. This study advances the state of the art by overcoming these hurdles and extending the
corpus-driven Cognitive research to morphology proper in an analysis of long and short forms of
predicative adjectives in Russian. The traditional grammars contrast the two forms stating that short
adjectives profile inherent characteristics and are more typical of formal registers where the long
form of adjectives profile more transient characteristics and tend to be more informal in use.
The study of morphological semantics presents two problems to the cognitive corpus linguist.
Firstly, it is harder isolate morphological meaning from lexical meaning than it is for syntactic
semantics. The large lexical variation typical of syntactic patterns makes constructional semantics
more straightforward to isolate. This study resolves this problem by including a lexical semantic
analysis and isolates the grammatical semantics by contrasting lexical - constructional pairings. In
effect, the grammatical semantics are identified by determining what remains unchanged when we
change the lexeme. Secondly, syntactic patterns often come in minimal pairs, or alternations, for
which es-tablished statistical techniques (Logistic Regression Analysis) have been developed.
These do not exist for looking at paradigms of several morphological variants. This study employs
Multiple Correspondence Analysis, an exploratory multivariate technique that indentifies patterns
of associations in the data even when there are more than two forms under investigation.
The data are taken from literature, news press, and online personal diaries. A total of 500
occurrences of the lexemes for happy and sad, in three forms - the short, the nominative long, and
the instrumental long form, are manually coded. The analysis covers semantic, extralinguistic, and
formal features. Grouping the nominative and instrumental forms of the adjectives together and
contrasting them with the short forms results in a binary alternation, permits a Logistic Regression
Analysis. The results of this analysis broadly confirm the hypothesis that there exists a difference in
degree of inherentness of characteristics expressed by the two forms. However, there are clearly
other variables affecting the distinction. Multiple Correspondence Analsyis is used to examine the
interaction of the three forms (nominative and instrumental sperately). The analyses reveal that for
one of the two lexemes, the instrumental long form, behaves more like a short form of the adjective.
It would seem, therefore that the traditional position is confirmed but with a qualification –
instrumentals can, at least for certain lexemes, profile a more inherent characteristic. Clearly further
research is needed, but the usefulness of Correspondence Analysis for the corpus study
morphological semantics is demonstrated.

Bibliography

Arppe, A. 2008. Univariate, bivariate and multivariate methods in corpus-based lex-icography - a


study of synonymy. Dept. of General Linguistics, University of Helsinki

33
Divjak, D. 2006. Delineating and Structuring Near-Synonyms. St. Gries & A. Stefa-nowitsch (eds),
Corpora in Cognitive Linguistics, 19-56. Berlin: Mouton.
Divjak, D. 2009. Mapping Between Domains. The Aspect-Modality Interaction in Russian. Russian
Linguistics, 33: 249-269.
Glynn, D. & Robinson, J. Forthc. Polysemy and Synonymy. Corpus methods and applications in
Cognitive Linguistics. Amsterdam: Benjamins.
Glynn, D. 2010. Synonymy, Lexical Fields, and Grammatical Constructions. A study in usage-
based Cognitive Semantics. H.-J. Schmid & S. Handl (eds). Cognitive Foundations of
Linguistic Usage-Patterns, 89-118. Berlin: Mouton.
Grondelaers, S., Geeraerts, D. and Speelman, D. 2007. A case for a cognitive corpus Linguistics. M.
Gonzalez-Marquez, et al. (eds), Methods in Cognitive Linguistics, 149-169. Amsterdam:
Benjamins.
Heylen, K. 2005. A Quantitative Corpus Study of German Word Order Variation. S. Kepser & M.
Reis (eds), Linguistic Evidence: Empirical, Theoretical and Computational
Perspectives, 241-264. Berlin: Mouton.
Janda, L. & Solovyev, V. 2009. What Constructional Profiles Reveal About Synonymy: A Case
Study of Russian Words for sadness and happiness. Cognitive Linguistics 20: 367-393.
Speelman, D. & Geeraerts, D. 2010. Causes for causatives: the case of Dutch doen and laten.
Linguistics of Causality. T. Sanders & E. Sweetser (eds). Cambridge: CUP.
Szmrecsanyi, B. 2010. The English genitive alternation in a cognitive sociolinguistics
perspective. D. Geeraerts & al. (eds), Advances in Cognitive Sociolinguistics, 141-166.
Berlin: Mouton.

34
Gorokhova Svetlana
St Petersburg State University

Syntactic overplanning in language production: Evidence from speech errors

This paper explores the effect of parallel processing of synonymous syntactic structures based
on an analysis of 104 naturally occurring Russian speech errors (slips of the tongue).
A well-known phenomenon resulting from parallel planning in natural language production is
a speech error known as the lexical blend (e.g. flown + driven → flivven), when two near-synonyms
(usually interpreted as contextual synonyms) are selected instead of one target lexeme. Word blends
are normally thought to occur when two lemma nodes are activated to an equal level, and both are
selected. Similarly, phrasal blends (e.g. if you’re not careful + if you don’t watch out→ if you’re
not watch out) are regarded as different formulations of the same message intertwined in speech
(Fay 1982; Coppock 2006), which can also be accounted for by the parallel selection of two
competing lemma nodes.
While semantic blends are well-known and have been observed by many researchers of
speech errors (Cutler, 1982; Fay, 1982; Garrett, 1980; Stemberger, 1990 etc.), little, if anything, has
been said of purely syntactic blends, i.e. blends of two synonymous syntactic structures with the
same set of lemmas.
An analysis of Russian speech error data reveals that such blends are not uncommon in
spoken Russian (a highly inflected language, where grammatical features such as case, number etc.
almost always surface as bound morphemes) and that, similarly to lemmas, syntactic structures do
compete during sentence production, e.g.

(1) Mozhno ja redisk-u doe-m + Mozhno mne redisk-u doe-st’ →


may I:NOM radish-ACC finish-SG.FUT may I:DAT radish-ACC finish-INF
Mozhno ja redisk-u doe-st’?
may I:NOM radish-ACC finish-INF
May I finish the radish?

(the error results from splicing the pronoun ja, the nominative case form of ‘I’, from the first phrase
and doest’, the infinitive form of ‘finish’, from the second. Both phrases are a form of request).

(2) On stal xud-oj + On stal xud-ym → On stal xud-om


he became thin-SG.NOM he became thin-SG.INS [non-existent word form]
He has grown thin.

(the error is caused by blending two different case forms, the nominative and the instrumental,
of the adjective xudoj ‘thin’. Either adjective case form collocates with the verb ‘become’).
Errors of this kind indicate that parallel planning is not confined to the lemma level of
sentence generation; instead, they suggest parallelism in syntactic planning at the positional level of
production (a level where, after the lemmas have been retrieved, the order of elements within a
phrase is specified and all inflectional processing takes place). The blends occur when the
processing mechanism fails to choose between the two competing positional level structures. Thus,
syntactic blends suggest that parallelism, or overplanning is a basic principle of language
production, which, alongside lexical retrieval, is involved in syntactic processing.
An implication for cognitive resource consumption is that syntactic structures are not
produced automatically; instead, syntactic blends indicate that generating a syntactic structure
requires significant processing resources (cf. Ferreira & Engelhardt, 2006).

35
Bibliography

Coppock, E. 2006. Alignment in Syntactic Blending. The State of the Art in Speech Error Research:
Proceedings of the LSA Institute Workshop. C. T. Schütze and V. S. Ferreira (eds.),
Cambridge, MA: MIT Working. Papers in Linguistics, 240-255.
Cutler, A. 1982. The reliability of speech error data. Slips of the Tongue and Language Production.
A. Cutler (ed.), Amsterdam: Walter de Gruyter/Mouton, 7-28.
Fay, D. 1982. Substitutions and splices: A study of sentence blends. Slips of the Tongue and
Language Production. A. Cutler (ed.), Amsterdam: Walter de Gruyter/Mouton, 717-
749.
Ferreira, F. & Engelhardt, P. 2006. Syntax and Production. Handbook of Psycholinguistics. M.
Traxler and M. A. Gernsbacher (eds.), London: Academic Press, 61-91.
Garrett, M. F. 1980. Levels of processing in sentence production. Language production, Volume I:
Speech and Talk. B. L. Butterworth (ed.), London: Academic Press, 177-220.
Stemberger, J. P. 1990. Word shape errors in language production. Cognition 35: 123–157.

36
Halder Guido
The University of Texas at Austin

A Construction Grammar Approach to German Support Verb Constructions

Support verb constructions (henceforth: SVCs) are constructions consisting of a verb with a
reduced meaning (when compared to the full verb) and a noun. For example, (1a) illustrates the full
verb meaning of German stellen ‘to put’, while (1b) uses stellen in a SVC.

(1) a. Peter stellt die Vase auf den Tisch.


‘Peter puts the vase on the table.’
b. Peter stellt die Frage zur Diskussion.
‘Peter brings the question up for discussion.’

Previous analyses (e.g. von Polenz 1963, Winhart 2002) provide a detailed account of the
function of the verb in SVCs. However, neither of the two approaches fully explains why certain
verb-noun combinations are unacceptable. Compare, e.g., geraten ‘to get into’, which can combine
with Brand ‘fire’ in (2a), but not with Feuer ‘fire’ in (2b), even though the two nouns are synonyms.

(2) a. Das Haus gerät in Brand.


‘The house catches fire.’
b. *Das Haus gerät in Feuer.
‘The house catches fire.’

Von Polenz (1963) does not explain why such selectional restrictions occur at all and Winhart
(2002) only provides an account for which arguments are obligatory, implicit, or optional.
This paper proposes a novel approach towards identifying selectional restrictions in German
support verb constructions by applying insights from Frame Semantics (Fillmore 1985) and
Construction Grammar. It differs from syntactic-centric and lexical-conceptual structure approaches
in that frame-semantic information is shown to influence directly a verb’s and a noun’s ability to
combine with each other. I argue that the nominalization in (2b) cannot combine with the support
verb because the frame-semantic information evoked by Feuer is incompatible with the frame
semantics of geraten. Thus, either the verb and/or the noun blocks the formation of a support verb
constructions. My analysis shows that in order for the support verb and the noun to be able to
combine, their frame-semantic information needs to be compatible (as in 1b). However, in some
circumstances such as in (2a, b), SVCs need to be listed as idioms in the lexicon because there does
not seem to be any compositional restrictions that allow geraten to combine with Brand ‘fire’ but
not Feuer ‘fire’. Based on a corpus of more than 300 SVCs with geraten I show that there are
different patterns of productivity and idomaticity that necessitate a network inheritance model.
Some SVCs, such as ins Rollen geraten ‘to start rolling’ allow widespread replacement of the noun
with near-synonyms. Other SVCs such as in (2a) do not allow such replacement. On this view, both
the abstract meaning of a SVC (e.g., in X geraten ‘to get into X’) and item-specific knowledge
needs to be captured to be able to account for the full range of SVCs headed by geraten. Therefore,
I posit a new construction which captures all the meanings expressed by SVCs with geraten.

37
Bibliography

Fillmore, C. J. 1985: Frames and the Semantics of Understanding. Quaderni di Semantica 6: 222-
254.
Von Polenz, P. 1963: Funktionsverben im heutigen Deutsch. Sprache der rationalisierten Welt.
Wirkendes Wort, Beiheft 5
Winhart, H. 2002: Funktionsverbgefüge im Deutschen zur Verbindung von Verben und
Nominalisierungen. Dissertation. Universität Tübingen.

38
Huumo Tuomas and Lehismets Kersten
University of Tartu
University of Tartu

Same conceptual content, different construals: On the Postposition / Preposition


variation of Finnish path adpositions

Finnish is a predominantly postpositional language which also has a few prepositions and
some two-faced adpositions that can be used in both functions (for a discussion, see e.g. Grünthal
2003). Many of such two-faced adpositions indicate a path, e.g., kautta 'via', läpi 'through', pitkin
'along', halki 'across' and yli 'over [dynamic]'. In the literature it has been pointed out that there are
differences between the variants that are related to their stylistical value: prepositions are more
literal or archaic whereas postpositions are neutral. However, this characterization seems to be valid
only in the spatial domain, since for instance temporal relations are usually more natural to be
expressed by prepositions. It has also been pointed out (Leino 1993, Salmi 1994) that there are
some semantic differences between the prepositional and postpositional variants of some Finnish
path adpositions, most clearly kautta 'via' and pitkin 'along'. These differences are, however, related
to the construal of the path than to the conceptual content of the expression (for the distinction, see
e.g. Langacker 2008). One crucial contrast goes between the meanings of actual motion (most
naturally expressed with postpositions but also by prepositions; example 1) and the directional
representation of static existence or occurrence of entities along the path (most naturally expressed
with a preposition, example 2).

1) Juoks+i+mme metsä+n läpi. ~ läpi metsä+n.


run+PST+1PL forest+GEN through through forest+GEN
'We ran through the forest.

2) Sien+i+ä kasv+o+i mon+in paiko+in läpi metsä+n.


mushroom+PL+PAR grow+PST+3SG many+PL.INS place+PL.INS through forest+GEN
'There were mushrooms growing in many places throughout the forest'.

This last-mentioned path type is also illustrated by Talmy's (2000:71) example There is a
house every now and then through the valley. According to Talmy, the example utilizes a sequential
perspectival mode, which involves a moving proximal perspective point with a local scope of
attention, i.e., a subjective scanning over a static configuration. In such an expression there is no
mover traversing the path, and the construal of the path is based on a subjective operation by the
conceptualizer. Our preliminary corpus analysis suggests that the usages of path postpositions
concentrate in examples indicating actual motion, whereas prepositional path expressions are more
flexible and heterogeneous in this respect – they can indicate actual motion (as in 1), but to indicate
the meaning like that of example 2, a preposition is needed. From the point of view of synonymy,
both variants thus indicate a path and convey a similar conceptual content, their difference lying in
the way of construing the path – a preposition is more flexible than a postposition and favors a
construal based on purely subjective operations.

Bibliography

Grünthal, Riho 2003. Grünthal, Riho 2003. Finnic adpositions and cases in change. Suomalais-
ugrilaisen seuran toimituksia 244. Helsinki: Suomalais-ugrilainen seura.

39
Langacker, Ronald W. 2008 Langacker, Ronald W. 2008. Cognitive grammar: A basic
introduction. New York: Oxford University Press.
Leino, Pentti 1993 Leino, Pentti 1993. Polysemia – kielen moniselitteisyys. Suomen kielen
kognitiivista kielioppia 1. Kieli 7. Helsinki: Helsingin yliopiston suomen kielen laitos.
Salmi, Tiina. 1994 Spatiaalinen ja metaforinen yli. – Leino, Pentti ja Tiina Onikki (toim.):
Näkökulmia polysemiaan: suomen kielen kognitiivista kielioppia 2. Kieli 8. Helsinki:
Helsingin yliopiston suomen kielen laitos, s. 164–188.
Talmy, Leonard 2000: Talmy, Leonard 2000. Toward a cognitive semantics. Volume 1: Concept
structuring systems. Cambridge: MIT Press.

40
Huumo Tuomas and Ojutkangas Krista
University of Tartu
University of Turku

Why do we need external and internal case forms for the gram sisä- in Finnish?

Finnish has a rich system of grammatical words, or grams, which have many different
functions. These functions extend from conceptually dependent adpositions to autonomous adverbs.
When used as adpositions, many grams take their complement in the genitive case, which reflects
the historical background of the construction: the grams were once nouns and the complements
were their genitive modifiers. This is why Finnish grams often have many local case forms in use,
many of them utilizing the case inflection to a remarkable extent.
The Finnish local case system has two paradigms: the internal cases (expressing notions like
‘inside’, ‘into’, ‘out of’), and the external cases (expressing notions like ‘at’, ’to the outside of’ and
‘from the outside of’). A pervasive feature in both systems is directionality: both case series have a
locative (‘in’/’at’) case, a lative (‘to’) case and a separative (‘from’) case. Some grams can be
productively inflected in all local cases. For instance, the stem sisä- ‘in’, has a full paradigm: sisä+llä
‘in’ [ADESSIVE] ~ sisä+ltä ‘from in’ [ABLATIVE] ~ sisä+lle ‘into’ [ALLATIVE] vs. sisä+ssä
‘in’ [INESSIVE] ~ sisä+stä ‘from in’ [ELATIVE] ~ sisä+än ‘into’ [ILLATIVE]. However, unlike
with proper nouns, it is not intuitively clear what the opposition between the internal and the
external case forms of this gram are. Both express, after all, a relationship of containment. In
contrast, the directionality opposition between different case forms of the gram is quite functional
and transparent.
In cognitive linguistics, it has been assumed that basic spatial concepts include relations such
as inclusion, separation, association and contact (Langacker 1987: 225–231). When applying such
notions to case-inflected grams, we need to remember that they express two different relations
simultaneously – one by the stem of the gram, another by the case ending. Another central
assumption of cognitive linguistics is that full synonymy does not exist. Thus internal vs. external
local case forms of a gram should not be synonymous either. However, the extensive literature on
grammaticalization has not paid much attention to the internal morphology of grams. Even in
typologically-based studies where the development of adpositions from nouns is described in detail
(e.g., Heine & al. 1991; Svorou 1993), the internal morphology of grams is discussed only
superficially if at all.
In our paper we study the spatial functions and actual usage of the gram sisä-, with special
reference to a) the internal division of labor between the stem and the case ending, and b) the
opposition between the internal vs. external case forms of the gram. We will show that in spite of
their apparent synonymy in simple examples, there are grammatical uses and kinds of landmarks
that are only compatible with one kind of a case form (internal or external) but not another. We will
also study the usage of the grams in spoken and written corpora, showing how the apparently
symmetric case paradigms are actually blended in usage: e.g., the lative meaning may be primarily
expressed by the internal case form of the gram, whereas the locative and separative meanings are
more often expressed by external case forms.

Bibliography

Heine, Bernd, Claudi, Ulrike & Hünnemeyer, Friederike 1991: Grammaticalization. A conceptual
framework. Chicago: The University of Chicago Press.
Langacker, Ronald W. 1987: Foundations of Cognitive Grammar. Vol. 1: Theoretical
Prerequisities. Berlin: Mouton de Gruyter.

41
Svorou, Soteria 1993: The grammar of space. Amsterdam: John Benjamins.

42
Häusler Sabine
Martin-Luther-Universität Halle-Wittenberg

Synonymy, sameness and semantic differences in verbal possessive constructions

The paper is dedicated to the problem of synonymy in syntax, i.e. whether competing
possessive constructions in the verbal phrase are supposed to be synonym in a way that they are
exchangeable. Working on possessive constructions a continuum between the advantage of
linguistic economy and the need of expressing connotations provides a rich field for research,
looking for the semantic overlap of these syntactic features.

The starting point of the investigation is a reconsideration of Heine´s quotation 1997:132 f.:
“On the one hand, there are languages where there is only one construction (e.g. the `have´-
construction in English), taking care of most or all of the spectrum of possessive notions, even if
there are other constructions in addition to cover parts of the spectrum. In Manding and Ewe,
however, the spectrum is divided up among different constructions. Whether this typological
contrast is suggestive of basic differences in the way possession is conceptualized, remains to be
investigated.”

Our analysis however is based on a corpus consisting of 16 languages of Europe both dead
(Gothic, Latin, Greek, Cl. Armenian, OCHS) and alive (Slavic and Baltic languages), using up to 5
different verbal possessive constructions each. All verbal possessive construction have been
classified – following Heine’s event-schemata – according to possessors, possessa, negation, special
semantic meaning in a given context, so the question of synonymy and disambiguation can be
answered both within a language, cross-linguistically and concerning historical developments
starting with PIE. So the study goes far deeper than the works on competing possessive cases
known within Classical Philology, by asking which possessive concepts are divided by non-
synonym constructions and whether there are implicational hierarchies to be found within the
disambiguation of these semantic groups.
A special consideration is dedicated to the question whether Indo-European languages use
different possessive construction to express the difference between alienable and inalienable
relations, what would contradict a concept of synonymy in possessive constructions. Although the
opinion expressed so far is a very sceptical one (cf. Bauer 2000: 164: “No Indo-European language
distinguishes explicitly between alienable and inalienable possession. Yet reinterpretation of
evidence from a number of daughter languages suggests that Indo-European at a very early stage
may have had this distinction.”), our answer is a positive one by giving evidence in a systematic
difference expressed by alternating constructions following the principles of grammaticalisation.

Bibliography

Bauer, B.L.M. 2000: Archaic Syntax in Indo-European, Berlin/New York.


Heine, B. 1997: Possession, Cognitive Sources, forces and grammaticalisation, Cambridge.
Seiler, Hj. 1983: Possession as an Operational Dimension of Language, Tübingen.
Seiler, Hj. 2008: POSSESSION: variation and invariance, in: Th. Stolz (ed.), Hansjakob Seiler.
Universality in Language beyond Grammar: Selected writings 1990-2007, Bochum.

43
Jürine Anni
University of Tartu

The Misspelling of Adpositional Phrases in Estonian –


Illiteracy or Manifestation of Semantic Difference?

In Estonian there is a noticeable tendency to misspell adpositional phrases. It is fairly


common to come across adpositional phrases written as one word, although according to the
Estonian orthography the adpositional phrases should always be written as two words. It is
hypothesized that the spelling issues of Estonian adpositional phrases are not only due to speakers’
poor knowledge of orthography, but are instead related to more basic tendencies operating in
languages, such as the ‘one form one meaning’ principle. Therefore, we consider the
orthographically correct analytic construction (Example 1a) and the misspelled synthetic
construction (Example 1b) as two competing constructions with potentially distinct meanings.

1. (a) Hoi-a dušš-I pea- Ø kohal


hold-imp.sg2 shower-sg.par head-gen over
‘Hold the shower head over your head’

(b) Su-l o-n nüüd katus *peakohal


you-sg.all be-sg3 now roof.nom over
‘You have a roof over your head now’

This view is in accordance with many cognitive linguistic accounts, such as Goldbergian
Construction Grammar, which refutes the possibility of full synonymy in a language, referring to
the principles of No Synonymy and Maximized Expressive Power. Thus, if we consider the two
constructions as syntactically distinct, there must be either semantic or pragmatic difference
between them (Goldberg 1995; Goldberg 2006). It is hypothesized that orthographical incorrectness
is related to the different meanings that these items may have for speakers – by ‘incorrect’
compounding speakers actually try to differentiate the abstract/metaphorical senses from the more
concrete ones. The results of a forced choice task – during which the participants were also asked to
explain their choice of construction in free form – conducted with 143 subjects confirmed this
hypothesis.
The first reason for considering distinct meanings of these constructions is the fact that the
orthographical issues arise when dealing with phrases that occur in metaphorical contexts (Example
1b) or express abstract functions, i.e. abstract relations in which motivation for using a particular
adpositional phrase is not transparent anymore (Example 2). So it seems that different linguistic
contexts trigger the use of different meanings: we may consider the adpositional phrases that occur
in various contexts as polysemous, furthermore the misspelled phrases are believed be lexicalizing
(Lehmann 2002; Brinton, Traugott 2005).

2. Ma ole-n selle-ᴓ peale mõel-nud.


I.nom be-sg1 this-gen on think-pst.ptcp
’I have thought about it’

It seems that another major factor triggering the variation is analogy, which is considered a
significant factor in language change (Aitchinson 2000: 216). As the studies I have conducted
(Jürine 2009) have shown, language users are not able to formulate any rules they might use to
make a selection between the two forms, but seem to rely on their feeling for language and use
44
analogies (they learn the orthography of some adpositional phrases and then extend the rule to
everything that seems comparable). The category of Estonian compound adverbs may set the
pattern to misspell. From this perspective, the current variation, i.e. spelling the Estonian
adpositional phrases as either one or two words, may indicate the beginnings of new compound
adverbs. In which context exactly the orthographically incorrect constructions are more likely to
occur is to be clarified via corpus-based research.

Bibliography

Aitchinson, Jean 2000. The Seeds of Speech: Language Origin and Evolution. Cambridge
University Press.
Brinton, Laurel J., Elizabeth Closs Traugott 2005. Lexicalization and Language Change.
Cambridge University Press.
Goldberg, Adele E. 1995. Constructions: a Construction Grammar Approach to Argument
Structure. Chicago: The University of Chicago Press.

Goldberg, Adele. 2006. Constructions at Work: The Nature of Generalization in Language. Oxford:
Oxford University Press.
Jürine, Anni 2009. Kas *metsavahel tähendab muud kui metsa vahel? [Does *metsavahel mean
something else than metsa vahel?] – Oma Keel, 18.
Lehmann, Christian 2002. New Reflections on Grammaticalization and Lexicalization. New
Reflections on Grammaticalization. Typological Studies in Language 49. Eds. Ilse
Wischer, Gabriele Diewald. Amsterdam; Philadelphia: John Benjamins Publishing
Company, pp. 1–18.

45
Klavan Jane
University of Tartu

Synonymous Locative Constructions in Estonian

In Estonian, we have among other synonymous items, several ways to express spatial
relations – we can use locative cases or adpositional constructions. The present paper builds on
previous research on this topic (Klavan et al. to appear) and the aim is to determine what does the
choice between the Estonian locative constructions with the adessive case (example 1a) and the
adposition peal (example 1b) depend on. The language data analysed comes from two linguistic
experiments – a production task and an acceptability task. The paper proceeds from the theoretical
premises of both Construction Grammar (Goldberg 1995, 2006) and Cognitive Grammar
(Langacker 1987, 2008), where one of the basic general assumptions is that of no-synonymy –
when two constructions differ syntactically, then they also differ either semantically or
pragmatically (Goldberg 1995: 67).

(1) a. vaas on laual


vase:NOM be-PRS:SG3 table:ADE
‘the vase is on the table’
b. vaas on laua peal
vase:NOM be-PRS:SG3 table:GEN on
‘the vase is on the table’

Previous research has shown (Klavan et al. to appear) that although the Estonian locative
cases and adpositional constructions are said to “express more or less the same meaning” in
traditional grammars of Estonian (Erelt et al. 1995: 34), there are differences in how language
speakers actually use these constructions. Klavan et al. (to appear) report the results of a forced
choice task and a production task carried out with 138 native speakers of Estonian, aged between 15
and 71. The results of these studies confirm that the type of relation between the Figure and Ground
plays a role: when the relation is abstract, the preferred choice is the adessive case; when we have a
spatial scene, where the relation is somewhat unusual or atypical (e.g. a book on top of an alarm-
clock), the language users prefer the adpositional construction. However, with common, everyday
spatial scenes (e.g. a vase on the table), the picture is not as clear-cut: both the adessive case and the
adpositional construction were used in the two studies with more or less the same frequency.
In order to determine what factors play a role in stereotypical spatial relations, two further
studies have been devised – a production task and an acceptability task. The production task tests
whether the following factors have an effect: animacy and size of the Figure, animacy and type of
the Ground, and the type of contact between the Figure and Ground. It is predicted, among other
things, that when the Ground is a thing (e.g. a tray), the preferred choice is the adposition peal
(‘on’) and when it is a place (e.g. a wall), the language users prefer the adessive case. Previous
research (e.g. Bartens 1978, Klavan et al. to appear, Ojutkangas 2008) shows that there might be an
effect of these factors, but in order to validate these claims, converging evidence needs to be
gathered.
Another group of factors that may influence the choice of these constructions with
stereotypical spatial relations are the syntactic ones. Although a small-scale corpus study conducted
by Rannat (1991) shows that the preference of either the synthetic or analytic form does not depend
on the syntactic composition of the clause (e.g. on the transitivity of the predicate verb, the type of
verb used, the clause element the locative phrase expresses, word order), these claims need to be
substantiated because her study is based only on her own intuition. In order to test whether there is

46
an effect of the syntactic composition in the choice between the Estonian adessive and the
adposition peal or not, an acceptability rating task has been devised.
Bibliography

Bartens, R. 1978. Synteettiset ja analyyttiset rakenteet lapin paikanilmauksissa. Suomalais-


ugrilaisen Seuran toimituksia 166. Helsinki: Suomalais-Ugrilainen Seura.
Erelt, M., Kasik, R., Metslang, H., Rajandi, H., Ross, K., Saari, H., Tael, K. and Vare, S. 1995.
Eesti keele grammatika I. Morfoloogia. Tallinn: Eesti Teaduste Akadeemia Eesti Keele
Instituut.
Goldberg, A. 1995. Constructions: A Construction Grammar Approach to Argument Structure.
USA: University of Chicago Press.
Goldberg, A. 2006. Constructions at Work: The Nature of Generalization in Language. Oxford:
Oxford University Press.
Klavan, J., Kesküla, K. and Ojava L. To appear. Synonymy in Grammar: The Estonian Adessive
Case and the Adposition peal ’on’. In S. Kittilä, K. Västi and J. Ylikoski (eds.), Studies
on Case, Animacy and Semantic Roles. John Benjamins.
Langacker, R. W. 1987. Foundations of Cognitive Grammar. Volume I: Theoretical Prerequisites.
Stanford: Stanford University Press.
Langacker, R. W. 2008. Cognitive Grammar. A Basic Introduction. Oxford: Oxford University
Press.
Ojutkangas, K. 2008. Mihin suomessa tarvitaan sisä-grammeja? Virittäjä 3, 382–400.
Rannat, R. 1991. Noomeni sünteetiliste ja analüütiliste vormide kasutus. [The Use of the Synthetic
and Analytic Forms of the Noun.] Unpublished BA thesis. University of Tartu, Institute
of Estonian and General Linguistics, Tartu.

47
Kholodilova Maria
Saint-Petersburg State University, hol_m@mail.ru

Relativization of subjects in Russian:


a case study of competition between syntactic synonyms

Russian uses two major strategies for relativizing on subjects, namely, active participle
strategy and relative pronoun strategy, pronoun kotoryj ‘which’ being by far the most frequent in
the latter case. In most of the cases in question the two structures are interchangeable without any
obvious difference in meaning (cf. (1a–b)).

(1a) Ja znaju aktrisu igrajuščuju v etom


I know actress playing in this
spektakle
stage.play
lit.: ‘I know the actress playing in this play’.
(1b) Ja znaju aktrisu kotoraja igraet v etom
I know actress who plays in this
spektakle
stage.play
‘I know the actress who plays in this play’.

The aim of the study was to specify the factors influencing the choice of relativization
strategy. The following parameters have been shown to be correlated with the choice of
relativization strategy: the mode of communication (oral vs. written); the tense, aspect and mood of
the verb in the relative clause; the presence of the reflexive suffix on the verb; the characteristics of
individual verbs as well as verb classes; the length of the relative clause; the presence of
complements of various types; negation; the type of the relative clause (restrictive vs. non-
restrictive); the syntactic position that the head of the relative clause occupies within the main
clause; intonational separation of the main clause and the relative clause. All the data were obtained
using the Russian National Corpus (ruscorpora.ru) and analysed with statistical methods.
It was found that there are a number of factors unequivocally determining the choice of
relativization strategy. These are partly due to restrictions on participle formation as there are no
participle forms inflected for conditional mood or future tense in standard Russian. Other
restrictions include the lack or rareness of some participle forms for certain verbs, such as byt’ ‘be’,
khotet’ ‘want’ and moč’ ‘be able’. When there is a need to express corresponding grammatical or
lexical meanings, kotoryj-clause is used almost invariably.
There are also some parameters that show strong statistical correlation with a particular
relativization strategy. For instance, kotoryj-clauses are more frequent in longer relative clauses in
which the verb has more dependents, especially if these are of clausal type (e.g. converbs).
Participles tend to appear in restrictive relative clauses. They are also much more common in
writing than in spoken language.
Both the absolute and statistical distinctions found in the study boil down to the following
generalization: if compared to kotoryj-clauses, participial clauses are more nominalized (in terms of
[Lehmann 1986]), that is, display more properties of a nominal (in the broader understanding of the
term) rather than a clause. For example, constraints on tense and constraints on possible
complements are both known to be characteristic of nominalized relative clauses
[Lehmann 1986: 671–672]. While some of the former are absolute in Russian and the latter only
show up in statistical correlations, there still seems to be a unifying macro factor behind both types
of constraints that influence the choice between the two syntactic strategies.

48
Bibliography

Lehmann, Сh. 1986. On the typology of relative clauses. Linguistics, 24 (4): 663–680.

49
Khoutyz Irina
Kuban State University

Synonyms in modern discourses: marking communicative choices

‘Synonymy, as a rule, is not complete equivalence’ (Jacobson 2002: 114): Roman Jacobson’s
statement stresses the thought that synonyms create linguistic variety, not sameness. The theory of
meaning views synonyms as the main manifestation of paradigmatic relationship (Кронгауз, 2002);
whereas the discourse analysis concentrates on the fact that the use of synonyms is the condition for
cohesion development which is achieved through words relating to each other within the text
(Cutting 2002: 2).
This paper, however, views synonymy as the increase of speakers’ communicative choices
which help them to express connotations and specify their attitudes to the topic of discussion. This
quality can be very well inferred from the following example: ‘Your clothes are too cool and you
have purple highlights.’ ‘Molichino, please!’ I cried. ‘Purple makes me sound like a … a teenager.
(Keyes 2008: 8). ‘Molichino’ is an Italian borrowing which means ‘purple’. Yet, as it is possible to
see from the example, it is not just an exact foreign equivalent. When this word is used in English
discourse, it acquires new connotations, shades of meaning. Thus, these adjectives can be described
as synonyms.
The expansion of synonymic fields is mostly affected by nonlinguistic causes and currently
illustrates the appearance of numerous borrowings which have been integrating into the language.
These new words often have local equivalents. However, the new words possess a strong pragmatic
core in their meaning (which is always at least slightly different from that of the local word) and
mark the discourse as contemporary and topical. The users of modern borrowings signal their
belonging to a world without boundaries, their willingness to be a part of common problems and
achievements (Khoutyz 2010). Moreover, speakers’ preferences to a certain synonym can
implicatively inform us about the genre of the discourse, its situational reference, speakers’ age,
profession, social status, etc.
This paper is going to observe how synonymy marks the discourse on many levels of meaning
transfer. To accomplish this task, first, I am going to discuss various approaches to studying
synonymy and dwell whether borrowings can be defined as full-fledged synonyms to local words.
Then, communicative properties such as modality, connotation, implicating, ideological preference,
etc. of a synonym are discussed. These properties are revealed through the analysis of functions of a
borrowed word in a contemporary discourse. In this case the use of a certain word serves as ‘the
conventional signal for a recurrent coordination problem’ (Croft 2000: 176). In conclusion, the
communicative choices expressed through the synonym use are classified.

Bibliography

Croft, W. (2000). Explaining Language Change. Oxford: Longman.


Cutting, J. (2002). Pragmatics and Discourse. London, New York: Routledge.
Jacobson, R. (2002). On Linguistic Aspects of Translation. In: The Translation Studies.
London,New York: Routledge, pp. 113-118.
Keyes, M. (2008). This Charming Man, Penguin Books, London.
Khoutyz, I. (2010). The pragmatics of anglicisms in modern Russian discourse. In: From
international to local English – and back again. Bern: Peter Lang, pp.197-208.
Кронгауз, М.А. (2002). Семантика. Москва: Российский государственный гуманитарный
университет.

50
Luraghi Silvia
Università di Pavia

How do languages deal with synonymy?

As stated in Taylor (2002:271), synonymy “tends to be avoided” as “an extravagant luxury”:


indeed, when words come to have the same meaning as a consequence of semantic change,
“‘corrective’ mechanisms come into play-one of the words may fall into disuse, or the words
become associated with different nuances, possibly of a stylistic or sociocultural nature.” (ib. 271-
2). In my paper I would like to show how such avoidance of synonymy has consequences on the use
of two prepositions, aná and katá, originally ‘upward’ and ‘downward’, in Ancient Greek. As early
as in the first written sources, these prepositions display a similar meaning extension. They both
mean ‘throughout’, ‘among’ in occurrences such as:

(1) Murmidónas d’ ár’ epoikhómenos thó#re#xen Akhilleùs


Myrmidon:ACC.PL PTC PTC walk:PART.PRS.M/P.NOM arm:AOR.3SG A.:NOM
pántas anà klisías
all:ACC.PL up hut:ACC.PL
“but Achilles went to and fro throughout the huts and let harness all the Myrmidons”
(Homer, Il. 16.155-156).
(2) ou mèn gàr pot’ áneu de#io#n ên, allà kat’ autoùs
NEG PTC PTC ever without enemy:GEN.PL be:IMPF.3SG but down DEM.ACC.PL
stro#phât’
range:IMPF.M/P.3SG
“for he was never away from the enemies, but ranged among them”
(Homer, Il. 13.556-557).

According to the traditional view, the semantic difference between aná and katá in such
occurrences lies in the fact that the former indicates that the trajector moves along a trajectory
which touches all relevant points in the landmark, while the latter indicates random motion touching
only part of the units included in the landmark (Eberling 1885, Luraghi 2003). More recently,
George (2006) has argued that the two prepositions had become complete synonyms, and that the
choice between the two only depended on their position in the verse. However, while it is true that
some of the occurrences mentioned by George do not point toward a difference in meaning, it is
also true that the distribution of the two preposition only partly depends on metric factors; in
addition, there is evidence from later prose (e.g. from Herodotus, cf. Luraghi 2003: 242) that the
meaning traditionally assumed for Homer was still preserved. The real difference between aná and
katá lies, as already indicated by Spitzner (1831), in the more restricted scope of aná with respect to
katá in Homeric Greek; later, aná became increasingly infrequent, as shown especially in Attic
prose, where the preposition only survives in some idiomatic expressions (Luraghi 2003: 227; see
further Bortone 2000). This development may be seen as an instance of ‘corrective mechanism’ in
the sense of Taylor (2002): in Homer, the two prepositions could often occur in the same contexts,
in which the semantic difference, even if existent, was irrelevant; this overlap turned into increasing
synonymy, and the use of aná became restricted to idioms.

Bibliography

Bortone, P. 2000: Aspects of the history of Greek prepositions. Oxford: Hilary Term.
George, C. 2006: The spatial use of aná and katá with the accusative in Homer. Glotta 82: 70-95
Ebeling, H. 1885: Lexicon homericum. Leipzig.

51
Luraghi, S. 2003: On the meaning of prepositions and cases: the expression of semantic roles in
Ancient Greek. Amsterdam: Benjamins.
Spitzner, F. 1831: Dissertatio de vi et usu praepositionum aná et katá apud Homerum. Wittenberg.
Taylor, J. R. 2002: Cognitive Grammar. Oxford: OUP.

52
Liu Dilin
University of Alabama

Construing and Conventional Usage in the Use of Synonymy: An Analysis of


Two sets of (Near)-synonymous Nouns Using Both Corpus and Elicited Data

Most corpus-based behavioral profile (BP) studies of (near)-synonyms so far have been on
verbs or adjectives (Divjak 2006; Divjak & Gries 2006; Gries 2001; Hanks, 1996; Liu, 2010). No
BP study appears to have examined synonymous nouns. Furthermore, corpus-based BP research,
though uniquely valuable, has the limitation of not being able to directly inquire why the producers
of the language data used a given linguistic item the way they did. The latter information may only
be obtained via solicited language, a type of data that recent research has shown to be valuable in
the study of synonymy (Arppe & Järvikivi, 2007; Geeraerts, 2010). Hence, this study examines two
sets of (near)-synonymous nouns (authority vs. power vs. right; duty vs. obligation vs.
responsibility), using both corpus and elicited data.
The study began with a BP analysis, using the Corpus of Contemporary American English
(COCA). It investigated the collocational/colligational patterns of the (near)-synonyms in each set,
including typical adjectives, post-nominal infinitives, and verbs used with the synonymous nouns.
The results including those of a multi-factorial test called “hierarchical configural frequency
analysis” (reported in tables below) indicate that pre-/post-nominal complements/modifiers were
effective in catching most of the semantic differences among synonymous nouns and in delineating
a “coarse” internal semantic structure of a synonymous-noun set. However, these examined
distributional patterns seemed unable to reveal some fine-grained differences that appeared to exist,
for the (near)-synonyms in each set, while displaying many different distributional patterns, also
exhibited some identical patterns, e.g. they sometimes took the same pre-/post-nominal
complements/modifiers (e.g. civic duty/obligation/responsibility; the authority/power/right to
declare war); it was not clear whether the different nouns used in the same distributional context in
each case had the same meaning.
To help answer this question, 32 of these difficult-to-distinguish uses in context were selected
from COCA to form a questionnaire (with the synonyms removed; see Appendix). 42 native
English speaking college students were asked to read each sentence and supply the missing word by
selecting from the three synonyms in the relevant set. The subjects were also asked to explain the
rationales for their choices. Statistical tests of the subjects’ choices (tabulated/reported in Appendix)
show a clear lack of consensus in 18 (56%) of the 32 items. Also, in 18 items, the choice by the
majority of the subjects differed from that used in the original COCA sentence. The analysis of the
rationales the subjects gave for their selections reveals that there were two major competing
motivating factors for their choices: 1) their construal of the situation (e.g. whether a subject
construed a given question as an issue of authority, power, or right) and 2) conventional usage
patterns (i.e. often a subject made a choice simply because it was “the idiomatic usage” he/she had
often heard). Many subjects were often torn between the competing factors in their decision
making. This new finding foregrounds the interface between construal and conventional usage as a
key issue in understanding synonymy, an issue that should be of interest to us, especially
cognitive/construction grammarians.

53
Bibliography

Arppe, A. (2008). Univariate, bivariate, and multivariate methods in corpus based lexicography: A
study of synonymy. Helsinki: University of Helsinki.
Arppe, A & Järvikivi, J. (2007). Every method counts: Combining corpus-based and experimental
evidence in the study of synonymy. Corpus Linguistics and Lingustic Theory, 3.2, 131–
159.
Charles, W. G. (2000). Contextual correlates of meaning. Applied Psycholinguistics, 21, 505-524.
Church, K. W., Gale, W., Hanks, P., & Hindle, R. (1991). Using statistics in lexical analysis. In U.
Zernik (Ed.), Lexical Acquisition: Exploring Online Resources to Build a Lexicon
(pp.115-164). Hillsdael, NJ: Lawrence Erlbaum.
Church, K. W., Gale, W., Hanks, P., & Hindle, R., & Moon, R. (1994). Lexical substitutability.In
B. T. S. Atkins and A. Zampolli (Eds.), Computational Approaches to the Lexicon
(pp.153-177). Oxford: Oxford University Press.
Divjak, D. (2006). Ways of intending: Delineating and structuring near synonyms. In S. ThGries
and A.Stefanowitsch (Eds.), Corpora in Cognitive Linguistics: Corpus-
basedApproaches to Syntax and Lexis (pp. 19-56.). Berlin and New York: Mouton de
Gruyter.
Divjak, D. & Gries, S. Th. (2006). Ways of trying in Russian: clustering behavioral profiles. Corpus
Linguistics and Linguistic Theory, 2, 23-60. The corpus of contemporary American
English Online. (2010). compiled and provided by Mark Davies of Brigham Young
University, available at http://www.americancorpus.org/.
Cruse, D. A. (1986). Lexical Semantics. Cambridge: Cambridge University Press.
Cruse, D. A. (2000). Meaning in Language: An Introduction to Semantics and Pragmatics.Oxford:
Oxford University Press.
Divjak, D. (2006). Ways of intending: Delineating and structuring near synonyms. In S. Th Gries
and A.Stefanowitsch (Eds.), Corpora in Cognitive Linguistics: Corpus-based
Approaches to Syntax and Lexis (pp. 19-56.). Berlin and New York: Mouton de
Gruyter.
Divjak, D. & Gries, S. Th. (2006). Ways of trying in Russian: clustering behavioral profiles. Corpus
Linguistics and Linguistic Theory, 2, 23-60.
Edmonds, P. & Hirst, G. (2002). Near synonyms and lexical choice. Computational Linguistics, 28,
105-144.
Firth, J. R. 1957. Papers in Linguistics, 1931–1951. New York: Oxford University Press.
Geeraerts, D. (1986). On necessary and sufficient conditions. Journal of Semantics, 5, 275-291.
Geerarerts, D. (2010). Theories of lexical semantics. Oxford: Oxford University Press.
Gries, S. Th. (2001). A corpus linguistic analysis of English –ic vs –ical adjectives. ICAME
Journal, 25, 65-108.
Gries, S. Th. (2004). HCFA 3.2. A program for R
Gries, S. Th. (in press). Statistics for Linguists with R: A Practical Introduction. Berlin: Mouton de
Gruyter.
Gries, S. Th. & Otani, N. (in press). Behavioral profiles: A corpus-based perspective on synonymy
and antonymy. ICAME Journal.
Hanks, P. (1996). Contextual dependency and lexical sets. International Journal of Corpus
Linguistics, 1 (1), 75-98.
Liu, D. (2010). Is it chief, main, major, primary, or principal concern? A corpus-based
behavioral profile study of the near-synonyms. International Journal of Corpus Linguistics, 15, 56-
87.
Miller, G. A. & Charles, W. G. (1991). Contextual correlates of semantic similarity. Language and
Cognitive Processes, 6 (1), 1-28.

54
Murphy, M. (2003). Semantic relations and the lexicon. Cambridge: Cambridge University Press.

55
Martí Solano Ramón and Raluca Nita
University of Limoges
University of Nantes

Metaphorical phrasal quantifiers and synonymy in a cross-linguistic perspective

Phrasal quantifiers or quantifying of-constructions such as a torrent of, mountains of, buckets
of, a barrage of, legions of, a sea of, etc. are numerous and tend to be regarded as synonymous. All
of these complex determiners fit into the same collocational framework (a) N1 + of + N2 and are
used to designate not only big quantities but also small ones as with a crumb of or a grain of. If
taken at face value, these quantifiers can be considered synonymous but once examined more
closely two different levels of synonymy become noticeable. Firstly, there is a broader level of
functional synonymy in which all of these constructions serve the same purpose, i.e. expressing
either a very large or a very small indeterminate quantity. And secondly, a narrower level of
conceptual synonymy in which only metaphorical expressions from the same source domain(s) can
be interchangeable. Thus, both armies of lawyers and legions of lawyers are attested, recurrent and
interchangeable but not *floods of lawyers, or a flood of information and a stream of information
but not *a legion of information. This paper addresses the major issue of contextual synonymy in
typological comparison. We claim that these quantifiers can be interchangeable in discourse only at
the second level of synonymy. We explore the actual uses of these expressions in a total of six
European languages, namely English, French, Italian, Portuguese, Romanian and Spanish. Our
claims are borne out by substantial corpus evidence of a large number of occurrences of this open
class of quantifiers. We argue that these conventional metaphorical patterns are shared by the afore-
mentioned languages and that they tend to collocate with the same nominal bases. Special attention
is drawn to the contrastive analysis of a selection of quantifiers which appear to be recurrent and
systematic in all six languages. We show that metaphorical phrasal quantifiers are, for the most part,
not language-specific as illustrated in the table below.
However, a detailed frequency analysis of the overall paradigm of quantifiers reveals
interesting information as to marked combinatory preferences across languages. In English a
barrage of criticism and a wave of criticism are far more frequent than a shower of criticism
whereas in French and in Romanian une pluie de critiques and o ploaie de critici are, respectively,
the preferred realisations. Spanish, Italian and Portuguese, in their turn, favour the metaphoricity of
words such as ola, oleada, onda, ondata, vaga, all of them meaning wave.

Bibliography

Croft W. 1990: Typology and Universals. Cambridge: Cambridge University Press.


Hanks P. 2006: Metaphoricity is gradable. Corpus-based Approaches to Metaphor and Metonymy,
A. Stefanowitsch & S. Th. Gries (eds), Berlin/New York: Mouton de Gruyter, 17-35.
Heine B. & Kuteva T. 2006: The Changing Languages of Europe. Oxford: Oxford University Press.
Lakoff G. & Johnson M. 1980: Metaphors We Live By. Chicago/London: The University of
Chicago Press.
Leclère Ch. & Brisbois-Leenhardt J. 2004: Synonymie de mots et synonymie de phrases: une
approche formelle. Lexique, Syntaxe et Lexique-Grammaire, Ch. Leclère, É. Laporte,
M. Piot & M. Silberztein (eds), Amsterdam/Philadelphia: John Benjamins, 389-404.
Renouf A. & Sinclair J. M. 1991: Collocational Frameworks in English. English Corpus
Linguistics, K. Aijmer & B. Altenberg (eds), New York: Longman, 128-143.

56
English French Italian Portuguese Romanian Spanish

A sea of Une mer de Un mare di Um mar de O mare de Un mar de


people gens gente gente oameni gente

An army of Une armée Un esercito Um exército O armata de Un ejército


lawyers d’avocats di avvocati de advogados avocati de abogados

A mountain Une Una Um monte de Un munte de Una montaña


of evidence montagne de montagna di provas dovezi de pruebas
preuves prove

57
Matuschek Michael and Gurevych Iryna
Technische Universität Darmstadt
Technische Universität Darmstadt

Beyond the Synset: Synonyms in


Collaboratively Constructed Semantic Resources

We present a comparative analysis of synonyms in collaboratively constructed and linguistic


lexical semantic resources and its implications for NLP research. Our focus is on the Wiki-based
resources constructed mostly by non-experts on the Web which rely on user collaboration for
quality management, as opposed to conventional sources of synonyms such as WordNet or thesauri.
The most prominent examples are Wikipedia 1 (a free Encyclopedia) and its dictionary spin-offs
Wiktionary 2 and OmegaWiki 3 , where the latter has a strong focus on crosslinguality. We will
examine three major ways how synonyms emerge in these resources, all of which imply a different
operational definition of synonymy. We will then show how these synonyms can be mined and used
building upon previous research in this field ((Zesch, Müller & Gurevych, 2008), (Wolf &
Gurevych, 2010)), and we will also examine what theoretical conclusions about the notion of
synonymy can possibly be drawn from our examinations.
The first part is the explicit encoding of synonymy, for example a link between word senses
in Wiktionary, where it can be argued that the user community agrees that they are synonymous;
this gives rise to a new notion of cognitive synonymy which is anchored in the “collective mind”.
Following earlier work considering German resources (Meyer & Gurevych, 2010), we analyze in
detail how synonyms are dealt with in different English resources, what problems arise for their
exploitation (e.g. due to inconsistencies) and how this compares to conventional lexical resources.
The second part is the implicit encoding of synonyms, e.g. deducing synonymy through a
transitive relation between two senses in Wiktionary. Another example is the redirect/link anchor
structure in Wikipedia. Here, it can be claimed that synonyms link to the same article. Contrary to
previous work (Nakayama et al., 2008), we show that this claim does not really hold, but some
interesting observations can be made regarding the link structure and how it relates to the idea of
synonymy. We also examine how links lead to insights about capital or subordinate traits of “distant
relatives” (cf. (Cruse, 1986)) which might give us a better idea of why words are perceived as
similar.
The third part does not rely on the structure of the resources but on the inference of synonymy
from context. Two examples are mining synonyms from the Wikipedia revision history (cf. (Nelken
& Yamangil, 2008)) as well as from an aligned corpus of Wikipedia and Simple Wikipedia. The
hypothesis is that (apart from spelling corrections) terms could be synonyms if they have been
replaced by each other in an article’s history (Kulessa, 2008) or if they are used interchangeably in
the “normal” and “simple” versions of an article (Zhu, Bernhard, & Gurevych 2010). Both
examples relate to the notion of propositional synonymy, but replacement of terms might also imply
that they were deemed invalid somehow; this observation could be another path for future research.
To substantiate our work, we give illustrative examples of synonyms collected from the
examined resources, and we provide statistical evidence about their structure and content.

1
http://www.wikipedia.org/
2
http://www.wiktionary.org/
3
http://www.omegawiki.org/

58
Bibliography

Cruse, D.A. 1986: Lexical Semantics. Cambridge Textbooks in Linguistics, Cambridge University
Press.
Kulessa, S. 2008: Mining Wikipedia's Revision History for Paraphrase Extraction (Master Thesis).
Technische Universität Darmstadt.
Meyer, C., & Gurevych, I. 2010: Worth its Weight in Gold or Yet Another Resource — A
Comparative Study of Wiktionary, OpenThesaurus and GermaNet. Proceedings of the
11th International Conference on Intelligent Text Processing and Computational
Linguistics, (pp. 38-49). Iaşi, Romania.
Nakayama, K., Pei, M., Erdmann, M., Ito, M., Shirakawa, M., Hara, T., 2008: Wikipedia Mining:
Wikipedia as a Corpus for Knowledge Extraction. Proceedings of Annual Wikipedia
Conference (Wikimania).
Nelken, R., & Yamangil, E. 2008: Mining Wikipedia's Article Revision History for Training
Computational Linguistics Algorithms. Proceedings of the Wikipedia and AI Workshop
at the AAAI Conference. Chicago, USA.
Wolf, E., & Gurevych, I. 2010: Expert-Built and Collaboratively Constructed Lexical Semantic
Resources for Natural Language Processing. Language and Linguistics Compass. (to
appear)
Zesch, T., Müller, C. & Gurevych, I. 2008: Extracting Lexical Semantic Knowledge from
Wikipedia and Wiktionary. Proceedings of the 6th International Conference on
Language Resources and Evaluation. Marrakech, Morocco.
Zhu, Z., Bernhard, D. & Gurevych, I. 2010: A Monolingual Tree-based Translation Model for
Sentence Simplification. Proceedings of The 23rd International Conference on
Computational Linguistics, Bei Jing, China. (to appear)

59
Montazeri Niloofar and Hobbs Jerry
University of Southern California
University of Southern California

Synonymy and Near-Synonymy in Deep Lexical Semantics

In an effort called “deep lexical semantics” (Hobbs, 2008), we characterize the meaning of
word senses in terms of axioms anchored in core theories that explicate basic concepts. For
example, a general senses of “give” can be defined by the axiom

(forall (x y z)
(iff (give-vb1 x y z)
(exist (e1 e2)
(and (cause x e1)(changeTo' e1 e2)(have' e2 z y)))))

That is, x gives y to z (in WordNet’s verb sense 1) if and only if x causes a change to a
situation in which z has y. The predicates cause and changeTo are explicated in a core theory of
events and event structure (cf. Hobbs, 2005), and the predicate have is explicated in Hobbs (2008).
(Our treatment of event verbs obviously owes much to the extensive body of work on lexical
decomposition that began with Gruber (1965).)
In this framework, the set of word senses of the same word normally constitutes a radial
structure, where the link between one word sense and the next is an incremental change in the
corresponding axiom. For example, the most general sense of “hit” in WordNet is sense 4, meaning
a change to being at a real or metaphorical location, as in “The thermometer hit 90 degrees.” The
less general sense 2 adds to sense 4 that there was a sudden impact, as in “The car hit a tree.” Sense
1 adds to sense 2 that the impact causes motion, as in “He hit the ball out of the park.”
This framework suggests a very natural characterization of synonymy and near-synonymy.
Synonymy occurs when two word senses of different words happen to be characterized by the same
axioms. Near-synonymy occurs when the associated axioms differ only by a proposition or two,
similarly to closely related word senses. For example, WordNet sense 1 for “provide” has the axiom

(forall (x y z)
(iff (provide-vb1 x y z)
(exist (e1 e2)
(and (cause x e1)(changeTo' e1 e2)(have' e2 z y)(need z y)))))

which differs from “give” (vb1) only in that what is provided is needed, whereas what is
given doesn't have to be. They are near-synonyms.
The axioms that explicate the meanings of words can capture subtle distinctions of meaning,
but only if the appropriate core theories have been explicated. In Edmunds and Hirst's (2002)
example, a blunder and a lapse are both errors but a blunder is caused by stupidity and a lapse by
neglect. Thus, we need to have explicated a commonsense theory of intelligence, attention and
other aspects of cognition (cf. Hobbs and Gordon, 2005). In the full paper we give more examples
of closely related nouns.
But the characterization of near-synonyms as differing only by a proposition or two may be
too broad. “raise” and “lower” differ only by a proposition; both mean to move something, but in
raising the direction is up and in lowering it is down. Similarly, “raise” and “rise” differ only in one
proposition, as raising is causing to rise. Are these pairs near-synonyms?

60
In the full paper we consider structural conditions on axioms that narrow the definition of
near-synonyms. But it is more likely that near-synonymy is not a natural kind about which people
have firm intuitions. Deep lexical semantics leads to a precise characterization of the relation
between two similar words, and this is probably more important than judging whether or not they
are near-synonyms.

Bibliography

Edmonds, Philip, and Graeme Hirst, 2002: Near-synonymy and lexical Choice. Computational
Linguistics Vol. 22, No. 2, pp. 105--144.
Gruber, Jeffery C., 1965. Studies in Lexical Relations, unpublished Ph.D. dissertation,
Massachusetts Institute of Technology, Cambridge, Massachusetts.
Hobbs, Jerry R., 2005: Toward a Useful Concept of Causality for Lexical Semantics. Journal of
Semantics, Vol. 22, pp. 181-209.
Hobbs, Jerry R., 2008: Deep Lexical Semantics. Proceedings of the 9th International Conference
on Intelligent Text Processing and Computational Linguistics (CICLing-2008), Haifa,
Israel, February 2008.
Hobbs, Jerry R., and Andrew S. Gordon, 2005: Encoding Knowledge of Commonsense
Psychology. Proceedings, 7th International Symposium on Logical Formalizations of
Commonsense Reasoning, Corfu, Greece, pp. 107-114, May 2005.

61
Mulli Juha
University of Eastern Finland at Joensuu

Observations on Synonymy and Lexical Variation of Body-part Idioms in


German

In general, synonymy is understood as referring to similarities of sense between lexical items.


However, the potential synonymy of idioms has been ignored in semantics, although it is
implicated, for instance, in the practice of cross-referencing in special idiom dictionaries
(Duden 2008, Schemann 1993). Despite extensive studies on the prototypical properties of idioms
in different languages, we can conclude that only little attention has been paid to potential
synonymy of idioms. Hence, the emphasis of the current literature suggests that synonymy is
restricted so single words (cf. Crystal 2003). In brief, many German studies have taken notice of the
ongoing large-scale electronic lexicography (e.g. DWDS-Korpus) in order to disclose contextual
conditions of idioms (cf. Hümmer 2004). Despite these and further studies, it has often been
emphasized that much more, for instance lexical information, is needed.
In general, lexemes denoting parts of the body (so-called body-part idioms) appear frequently
in German idioms. According to Heringer (2004: 175), approximately 20 per cent of the idioms in
German contain a body-part noun as their constituent, e.g. einen kühlen Kopf bewahren ‘to keep a
cool head’. Moreover, idioms that contain (at least) one body-part noun, account for the biggest
group of German idioms in general (Schemann 1993). First, prototypical idioms are not only
syntactically but also lexically restricted expressions containing more than one word. Second,
idioms are per definitionem non-compositional, i.e. the meaning of an idiom cannot be derived from
the meanings of its components. In addition, idioms are institutionalized, i.e. they are known to
native speakers of the language and can be found in dictionaries (Nunberg et al. 1994). Nonetheless,
there have also been attempts to make idioms relatively flexible (cf. Penttilä 2006) and studies
related to these aspects have shown that idioms include more variation than assumed hitherto and
much of the variation is systematic (Fellbaum 2007). The presents study has three major aims. First,
lexical variation of German body-part idioms will be examined. The ultimate goal is to find out
whether the testing of native-speakers of German could provide novel aspects of the variation of
body-part idioms. Second, the crucial question to be answered is: Is there only lexical variation, or,
in general, should we consider this as synonymy? In other words, the present study sets out to see
whether synonymy and lexical variation are measurable properties of body-part idioms in German.
In brief, this paper tries to shed some light on the question where to draw the line between
synonymy and lexical variation in German idioms.

Bibliography

Crystal, David 2003. A Dictionary of Linguistics and Phonetics. Malden, MA: Blackwell Pub.
Duden. Redewendungen. Wörterbuch der deutschen Idiomatik 2008. Herausgegeben von der
Dudenredaktion. 3., überarbeitete und aktualisierte Auflage. (Der Duden in 12 Bänden.
11). Mannheim: Dudenverlag.
DWDS-Korpus: http://www.dwds.de
Fellbaum, Christiane (ed.) 2007. Idioms and Collocations: Corpus-based Linguistic and
Lexicographic Studies. London: Continuum.
Heringer, Hans Jürgen 2004. Interkulturelle Kommunikation. Grundlagen und Konzepte. 2.,
durchgesehene Aufl. (UTB 2550: Sprachwissenschaften). Tübingen und Basel:

62
Francke.Hümmer, Christiane 2004. A Corpus Based Approach to Near Synonymy of German
Multi-Word Expressions. In: Petr Sojka, Karel Pala & Pavel Smrz (eds.) Proceedings of
the Second International WordNet Conference. Brno: Masaryk University.
Nunberg, Geoffrey, Ivan A. Sag & Thomas Wasow. 1994. Idioms. Language, 70, 3, 491-538.
Penttilä, Esa 2006. It Takes an Age to Do a Chomsky: Idiomaticity and Verb Phrase Constructions
in English. PhD thesis, University of Joensuu.
Schemann, Hans 1993. Deutsche Idiomatik. Die deutschen Redewendungen im Kontext. Stuttgart:
Klett.

63
Nacey Susan and Egan Thomas
Hedmark University College
Hedmark University College

The (near-)synonyms begin and start: evidence from translation corpora


Non-finite complement constructions with the English matrix verbs begin and start exhibit a
degree of similarity, if not outright synonymy, along two axes. On the one hand there is the contrast
between the two matrix verbs. As Dixon puts it, “In many sentences start and begin may be
substituted one for the other with little or no change in meaning” (Dixon 2005: 181). On the other
hand, it is often difficult to discern substantive differences in meaning between pairs of
constructions containing the same matrix verb but differing in choice of complement form. Thus
Quirk et al. (1985: 1192), commenting on the choice of to- infinitive or -ing complements, note that
in the case of many examples “there is no observable difference of meaning between the
constructions”. Freed also underlines the similarity between the various constructions in certain
contexts, writing “there are cases or contexts in which the difference between them does not matter”
(Freed 1979: 75). This impression of mutual entailment is also emphasised by Egan (2008), who
writes “There are many cases where a begin construction may be seen to entail a start construction,
a to infinitive construction an -ing construction, and vice versa (i.e. there are many utterances of the
type ‘she began to do X’ which, if true, guarantee the truth of ‘she started to do/doing X’)” (Egan
2008: 256).
Although all four of the authors mentioned above stress the mutual substitutability of the four
constructions in certain contexts, they also all maintain that there exist subtle distinctions between
them, in other words contexts in which the substitution of one for the other would lead to
differences in interpretation. There is, however, considerable disagreement between these and other
authors as to the exact nature of the relevant distinctions (see also, for example, Mair 2003).
In this paper we attempt to pin down more closely the extent and nature of the similarities and
differences between the various constructions containing begin and start using translation corpora, a
source of evidence which is as far as we know not previously mined. We look at how constructions
containing begin and start are translated into Norwegian, a language which contains the cognate
verbs begynne and starte. Our data for this part of the study come from the English-Norwegian
Parallel Corpus (see Johansson 2007), which contains 426 tokens of begin and 277 of start. We
then use the Oslo Multilingual Corpus to examine translation equivalents of constructions
containing the Norwegian matrix verbs begynne and starte into French, German and English. We
investigate the extent to which cross-linguistic similarities and differences in choice of translation
options may mirror putative lexical, constructional, functional and formal similarities and
differences in the original expressions. In so doing, we utilize translation corpora to shed light on
the degree to which begin and start may be considered synonymous.

Bibliography

Dixon, R. M. W. (2005). A semantic approach to English grammar (2nd. ed.). Oxford: Oxford
University Press.
Egan, T. (2008). Non-finite complementation: a usage-based study of infinitive and -ing clauses in
English. Amsterdam: Rodopi.
Freed, A. F. (1979). The Semantics of English aspectual complementation. Dordrecht: Reidel.
Johansson, S. (2007). Seeing through Multilingual Corpora : On the use of corpora in contrastive
studies. Amsterdam: John Benjamins.
Mair, C. (2003). Gerundial complements after begin and start: Grammatical and sociolinguistic
factcors, and how they work against each other. In B. Mondorf & G. Rohdenburg

64
(Eds.), Determinants of grammatical variation in English (pp. 329-345). Berlin:
Mouton de Gruyter.
Quirk, R., Sidney Greenbaum, Geoffrey Leech & Jan Svartvik (1985). A Comprehensive grammar
of the English language. London: Longman.

65
Nissilä Niina and Pilke Nina
University of Vaasa
University of Vaasa

Synonymy in specialized communication – a terminological approach

Specialized communication sets different requirements on the language than everyday


communication. Language for specialized purposes (LSP) is a tool for classifying, describing and
mediating information within different subject fields. The language used in LSP communication
should thus be characterized by unambiguousness, precision and logicality. The ideal situation for
achieving precise and unambiguous communication is monosemy, i.e. one concept is represented by
only one term. But it is obvious that polysemy, homonymy and synonymy have a significant role
also in the communication between experts within the same subject field in a similar way as
linguistic variants in general language. Existence of synonymous expressions is especially
combined with fields where the development is fast and where overall coordination still is missing.
In this paper we first focus our attention on how synonymy is discussed in terminological
literature. Terminology science is a field of study of concepts and terms based on systematic
concept analysis. In general synonymy is seen as an unwanted phenomenon in terminology because
it can prevent mutual understanding within the subject field. (Arntz & Picht 1989). There seem to
exist somewhat different views among the researchers within the field of terminology. According to
Lotte (1993) synonymy is actually dangerous: “Die Synonymie von Termini ist auch gefährlig, und
sie sie muß eliminiert werden, d.h. jedem Bergriff darf nur ein Terminus zugeordnet werden”. Irgl
(1989) represents another view by pointing out that synonymy is caused by objective and subjective
factors. This can be seen e.g. in the language of economics, where synonymy is surprisingly
common. The developer of the General theory of terminology in the 1930s, Wüster (1991), states
that an absolute requirement on precision is a theoretical ideal and gives a categorization of
synonymies based on meaning (Sachbedeutung, Mitbedeutung, Sach- und Mitbedeutung) and
uniformity (Spalt- und Systemsynonyme).
Secondly we analyze and categorize different kinds of synonymous expressions in
vocabularies compiled using terminological methods. We intend to examine the term structures and
the order of terms in terminological records including one or more qualified synonymic expression
to the entry term both within and across languages. The synonymous terms in one language are
studied also in relation to the aim or focus in the actual context due to so called dimensions (cf.
Bowker1997) The terms entrance floor and ground floor, for example, represent the same concept
but the choice between them depends on the point of view (entrance to the building vs. number of
stocks). The analysis of the definitions in the vocabularies is based on a comparison between the
terms (entry term, synonym/s) and the essential and delimiting characteristics in the description
including the note section.

Bibliography

Arntz, R. & H. Picht (1989). Einführung in die Terminologiearbeit. Hildesheim etc.: Georg Olms
Verlag.
Bowker, L. (1997). Your say “flatbed colour scanner” and I say “colour flatbed scanner” A
descriptive study of the influence of multidimensionality on term formation and use
with special reference to the subject field of optical scanning technology. In:
Terminology: international journal of theoretical and applied issues in specialized
communication vol 4 (2), 275–302. Amsterdam: Benjamins.

66
Irgl, V. (1989). Synonymy in the language of business and economics. In: Special language: from
humans thinking to thinking machines. 275–282. Eds. Laurén Ch. & M. Nordman.
Clevedon, Philadelphia: Multilingual Matters.
Lotte, D.S. (1993). Aufgaben und Methoden zur Regelung von technischer Terminologie. In:
Ausgewählte Texte zur Terminologie. 190–220. Hg. Laurén, Ch. & M. Nordman. Wien:
Termnet.
Wüster, E. (1991). Einführung in die Allgemeine Terminologielehre und Terminologische
Lexikographie. 3. Auflage. Bonn: Romanistischer Verlag.

67
Nedjalkov Igor
St-Petersburg State University

Translation variants as Another Type of Synonymy

Synonymy is a very important cognitive concept for describing a language lexicon and
grammar both from the systemic-semantic and functional-pragmatic aspects since it enables
linguists to detect minor semantic differences and combinability peculiarities of semantically
similar words and constructions (cf. quick – fast – rapid – swift; The hunter killed the bear – The
bear was killed by the hunter; He always comes to work in time – He is never late for his work;
John bought the book from Mary - Mary sold the book to John) and, consequently, to find out
common and differentiating contexts of semantically similar words and constructions. A broader
perspective in approaching the concept of synonymy can be also achieved by including in the
domain of synonymy analysis several translation variants of one and the same original text. The
paper deals with the main types of English translation variants (both phrasal and clausal) of Russian
fiction written by Fyodor Dostoyevsky, Leo Tolstoy, Anton Chekhov and Mikhail Bulgakov. These
translation variants may be regarded synonymous since they convey one and the same meaning of
the original text. Selected passages of the original texts contain non-trivial (either polysemous or
occasional) words and phrases, as the verbs in the examples below:

(1) On blagopolučno izbegnul vstreči s svojeju hoz’ajkoj na lestnice (Dostoyevsky. Crime and punishment) –
a. He had successfully avoided meeting his landlady on the staircase (C. Garnett);
b. He was lucky to avoid a meeting with his landlady on the stairs (D.Magarshack);
c. He had been lucky enough to escape meeting his landlady on the staircase (J.Katzer);
d. He had succeeded in avoiding an encounter with his landlady on the stairs (McDuff);
e. He was fortunate enough not to meet his landlady on the stairs (Anonymous translation; London
1997);
(2) Vs’o smeshalos’ v dome Oblonskih (Leo Tolstoy. Anna Karenina) –
a. Everything was in confusion in the Oblonskys’ house (C. Garnett);
b. Everything was in confusion in the Oblonskys’ household (D.Magarshack);
c. All was confusion in the Oblonskys’ house (Pevear / Volokhonsky);
d. Everything had gone wrong in the Oblonsky household (R. Edmonds);
e. Everything was upset in the Oblonskys’ house (L. and A. Maude);
f. The Oblonsky home was in turmoil (M. Wettlin);
g. Everything at the Oblonskys’ was topsy-turvy (J. Carmichael).

The analysis of semantically identical or similar translation variants of the types presented
above allows to take into consideration phrases and clauses expressing in different ways the same
meaning of the original text and containing words which express similar concepts but belong to
different parts of speech (cf. a Russian adverb blagopolučno and an English verb succeed in
example (1) and a Russian verb smeshalos’ and an English noun confusion in example (2)).
The second part of the paper analyses different English translations of Russian words
expressing emotions with special attention to cases when original Russian words are rendered by
the English corresponding lexemes belonging to another part of speech. The analysis included 320
Russian sentences and phrases of five Chekhov’s plays (including stage directions), and since the
number of English translations equaled five to six, the quantity of English translation variants was
not less than 1650 sentences and phrases. In 110 English translations (approximately 7 % of all the
translations) various changes in part-of-speech status of the original lexemes were registered. It
should be noted that all twelve theoretically possible changes of part-of-speech status of the original
Russian lexemes were present in 110 English translations: 1) Noun in the Russian text (RT) à
Verbal form in the English translation (ET) (8 cases), 2) Noun in RT -> Adjective in ET (14 cases),
3) Noun in RT -> Adverb in ET (13 cases), 4) Verb in RT -> Noun in ET (6 cases), 5) Verb in RT -
68
> Adjective in ET (9 cases), 6) Verb in RT -> Adverb in ET (6 cases), 7) Adjective in RT -> Noun
in ET (3 cases), 8) Adjective in RT -> Verb in ET (2 cases), 9) Adjective in RT -> Adverb in ET (7
cases), 10) Adverb in RT -> Noun in ET (13 cases), 11) Adverb in RT -> Verb in ET (10 cases), 12)
Adverb in RT -> Adjective in ET (19 cases). It turns out that Russian adjectives are most stable
(only 12 cases of part-of-speech status change), whereas Russian adverbs are most changeable (42
cases of part-of-speech status change, or about 38 % of all cases involving part-of-speech change).
So, English words expressing the same emotive concepts but belonging to different parts of speech
can be regarded as either synonymous or quasi-synonymous.

69
O’Connor Kathleen and Corteel Céline
Universités de Lille 3
Université d’Artois

Autre – different vs. other – different: A contrastive approach

In French, the relational adjectives autre ‘other’ and différent ‘different’ are binary predicates
that establish a relationship of ‘difference’ between two entities (e.g. Noailly, 1999; Schnedecker,
2002). As might be expected, the two terms can sometimes be used interchangeably with very little,
if any, contrast in meaning:

(1) Marie est allée au Louvre, mais Pierre a préféré visiter un (autre) musée (différent).
Mary went to the Louvre, but Pierre wanted to visit (another) (a different) museum.

Moreover, both adjectives establish a relationship between the phrase containing them and a
second element elsewhere in the context (e.g. Van Peteghem, 1997), for example in their
complement or to their left. In example (1), one of the arguments is part of the phrase containing the
adjective (i.e. musée), whereas the second argument is to the left of this phrase (i.e. Louvre). In (2),
the second argument is found in the complement to the phrase containing the adjective:

(2) J’ai choisi un sac différent du tien/un autre sac que toi.
I have chosen a different purse from yours/another purse than you.

These similarities can be seen to account for the fact that, in monolingual French dictionaries
(e.g. Le grand Robert or the Trésor de la langue française), the entries for each term refer readers to
the other.
However, sentences such as (1) where the two terms can be used interchangeably are in fact
rare. There are a number of contexts in which only one of the two terms is possible without a
change in interpretation:

(3) Son mégot à peine écrasé, il alluma une autre cigarette/*une cigarette différente.
He had scarcely put out his cigarette when he lit another/a different one.

(4) J’ai acheté trois robes différentes/#trois autres robes pour partir en vacances.
I bought three different dresses/three other dresses for my vacation.

This brings us to the goal of the present paper. Using constructed and attested examples, we
will describe the semantic and referential properties of autre and différent on the basis of the
different possibilities for syntactically encoding their arguments. More specifically, it will be shown
that différent functions as a symmetrical adjective with a reciprocal sense whose base meaning
allows it to act as a determiner. Autre, on the other hand, appears to function more as a comparative
than as a typical adjective. It has a negative meaning in that it excludes a given referent as a
potential referent for the phrase in which it appears.
These conclusions will then be compared with the data concerning the English adjectives
other and different. While superficially similar, both semantically and syntactically, the uses of the
English adjectives do not line up perfectly with the French data, as shown in (5):

70
(5) J’ai choisi un autre sac que toi.
*I have chosen another purse than you.

Thus, in English, it appears that the distribution of other is more restrained than in French,
revealing that the terms are not synonymous cross-linguistically.

Bibliography

Breban, T. 2010. English adjectives of comparison: Lexical and grammaticalized uses (Topics in
English Linguistics). Berlin: Mouton de Gruyter.
Dotlačil, J. 2010. Anaphora and Distributivity. A study of same, different, reciprocals and others.
LOT dissertation series 23.
Noailly, M. 1999. L’adjectif en français. Paris: Ophrys.
Schnedecker, C. 2002. Présentation: Les adjectifs « inclassables »: des adjectifs du troisième
groupe? Langue française 136: 3-19.
Van Peteghem, M. 1997. Mécanismes anaphoriques sous-jacents aux ‘indéfinis’ autre et même. In
W. DeMulder, L. Tasmowski-DeRyck and C. Vetters (eds), Relations anaphoriques et
(in)coherence (pp. 187-200). Amsterdam: Rodopi.

71
Oversteegen Eleonore
Tilburg University

What ‘s in a synonym?

Synonymy lies at the heart of (lexical) semantics. The interest in this phenomenon, as visible
in the history of linguistics, can be explained by the fact that insights in the equality or similarity of
meaning may shed light on meaning itself. Nowadays, it is possible to obtain huge quantities of
(NLP) applicable information about word usage by analyzing large data collections. This
information, however, does not necessarily help us understand what is required, in terms of
linguistic theory, for two words to be (nearly) synonymous. In our research, we adopt an enriched
version of a decompositional approach to word meaning. Our main research question concerns the
relative influence of six well defined word properties on (experimentally obtained) synonymy
judgements.
In our research, we define a set of factors relevant to word meaning. These factors, envisaged
as word properties, are not novel. In some form, they were introduced in studies of a syntactic,
semantic or pragmatic nature. The composition of these factors in a instrument for measuring
synonymy, however, is original. Three of them are coined paradigmatic, since they can be seen as a
choice from a set of alternatives. The paradigmatic ones are: conceptual properties, register, and
reading. The three others are labelled syntagmatic, since they refer to some structural pattern:
selection restrictions, telic role, and agentive role (cf. Pustejovsky, 1995).
The theoretical part: A set of candidate word pairs is selected, existing of more or less
synonymous words. For each word property, a value is attributed to the candidate word pairs,
reflecting the degree of similarity of the pair. In order to warrant intersubjectivity in the analyses, a
reliable dictionary of synonyms is used (…). For each word pair, the combined similarity scores for
all word properties is defined as the score on semantic distance. This score ranges from zero (strong
synonymy) to six (weak synonymy).
The experimental part: In three experiments, the correlation has been evaluated between
criteria and distance score on the one hand, and judgements by native speakers on the other hand. In
designing these experiments, special attention was given to three methodological issues: the
selection of the set of word pairs, the linguistic context in which the words were to be presented,
and the exact phrasings of the judgements asked for.
Judgements of synonymy appeared to be closely linked to the syntagmatic properties of
words, with selection restrictions as the most determinative factor. This factor was followed by the
syntagmatic properties telic role and agentive role. The paradigmatic properties played a
surprisingly modest role.
Evidently, in deciding on synonymy, it’s the deeper layers of linguistic knowledge about
structural patterns that are consulted. The results seem to agree with the current view (e.g.
Jackendoff, 2007), that the borderline between grammar and lexicon is not as rigid as we used to
think Moreover, the results engender a new set of questions about word meaning and the human
mind. Why does the pragmatic property register play such a trivial role? What makes the telic role
so important? Why is it more important than the agentive role?

72
Palolahti Maria
University of Helsinki

Re-thinking synonymy: a cognitive science perspective

The goal of my talk is to give a cognitive science perspective on some founding questions
concerning the nature of synonymy, including the cognitive definition of synonymy, its potential
usefulness as a theoretical concept in cognitive science, and translatability (similarity of meaning)
across languages. My theoretical starting points are the philosophical notions of synonymy, the
holistic view of meaning, and the indeterminacy of translation thesis ('the Gavagai argument')
formulated by the philosopher W.V.O. Quine (1951, 1960). I will first briefly present these points,
after which I will introduce some (partial) objections to these views. The objections also include
two examples from current empirical language research where the notions of lexical as well as
constructional synonymy would seem to play a theoretically important role.
The first example comes from the scientific study of bilingualism, where the theoretical
concepts of synonymy and/or translational equivalence could be seen as potentially foundational.
Consider, for example, the English words pizza / boy / cognitive science and their Finnish
translation equivalents pitsa / poika / kognitiotiede and imagine then a fully bilingual speaker of
English and Finnish. When words from two different languages are compared with each other this
way, no usage-based techniques can be applied for identifying similarity of word meaning.
However, Francis (2005) has argued, that for translation equivalents in general, the current
empirical evidence from psycholinguistic studies on bilingualism strongly favours the scientific
hypothesis of shared semantic systems and shared semantic/conceptual representations.
The second example comes from cognitive neuroscience. Consider the following Finnish
sentences (and their English counterparts):

(1) Poja-t osta-isi-vat pitsa-t huomenna


boy-NOM.PL buy-COND-3PL pizza-ACC.PL tomorrow
'(The) boys would buy (the) pizzas tomorrow'
(2) Pitsa-t oste-tta-isi-in huomenna poikien toimesta
pizza-ACC.PL buy-PASS-COND-AGR tomorrow by the boys
'(The) pizzas would be bought tomorrow by (the) boys'.

Sentence (2) is an example of the Finnish morphological (‘impersonal’) passive construction.


The passivisation in Finnish is a morphosyntactic valency-changing operation, where the initial
subject NP of the active sentence is either demoted or deleted, and the initial object NP (if there is
any) is moved to sentence-initial position. At the same time the finite verb undergoes major
morphological changes, including insertion of the passive marker -(t)ta and the suffix -Vn, which is
a special grammatical person marker that appears in finite passive constructions. In a recent
experimental study (Palolahti et al., in preparation) real-time semantic and grammatical processing
of active and passive Finnish sentences were studied using event-related potentials (ERPs). The
obtained results were then compared to the results of a previous ERP study by Kim & Osterhout
(2005), where English periphrastic passives and active English sentences were used as experimental
stimuli. Interestingly, the ERP results from these two studies were observed to be practically
identical, regardless of the differences between languages and particular linguistic constructions in
question.

73
Bibliography

Kim, A. & Osterhout, L. (2005). The independence of combinatory semantic processing: Evidence
from event-related potentials. Journal of Memory and Language, 52, 205-225.
Francis, W.S. (2005). Bilingual semantic and conceptual representation. Handbook of bilingualism:
psycholinguistic approaches. J. F. Kroll & A. M. B. de Groot (eds.). Oxford University
Press.
Quine, W.V.O. (1951). ‘Two dogmas of empiricism’, The Philosophical Review, 60, January 1951.
Quine, W.V.O. (1960). Word & Object, Cambridge: M.I.T. Press.

74
Paulsen Geda
Åbo Akademi University

Resemblance and variation of causatives expressing social relations

In my paper, I discuss closely related languages Estonian and Finnish and their ways to
encode and conceptualize social dominance. By social dominance I mean dominance and causation
between human actors. There are both similar and different linguistic means in Estonian and
Finnish for expressing social causation – synthetic and analytic. From the typological point of view,
morphological causative derivation is not found in all languages. The morphological causatives
derived with the suffix ttA are a widespread and idiomatic means of expression in Finnish;
compared to that, the use of causative derivation with the corresponding morpheme ta is in Estonian
estimated to be more restricted (Kasik 1989, 2001). However, some of the morphological causatives
are used in rather similar contexts in these languages, compare the examples (1-2):
(1) Hallitus juoksuttaa maanviljeljää (Fi.)
Valitsus jooksutab põllumeest (Est.)
government-nom run-cause-3sg farmer-part
‘The government is running the farmers around’

(2a) Liverpool kyykytti Chelsean 3:1 (Fi.)


Liverpool-nom squat-caus-pst-3sg Chelsea-acc 3:1
‘Liverpool beat Chelsea 3:1’
(2b) Liverpool seljatas Chelsea 3:1 (Est.)
Liverpool-nom back-caus-pst-3sg Chelsea-acc 3:1
‘Liverpool beat Chelsea 3:1’

My presentation will focus on the morphological causatives expressing social dominance in


these languages and leave the lexical and analytic outside the scope. How are social dominance
relations encoded in these languages, what are the similarities and differences? How to capture the
relations encoded between humans and the social hierarchies? What is the role of responsible
conventional behaviour in causative constructions? I argue that causative verbs can have specialized
uses where the expressions of social relations and attitudes have important roles. I approach the
derivatives within the framework of conceptual semantics and argue for a prototype semantic
interpretation for these derivatives. I suggest that the abstract type of prototype templates can be
used for comparison of the verbs as well as for revealing the constructional extension and
idiosyncrasy. I assume that there are partly overlapping prototype-based patterns and particular
causative constructional patterns in these languages.

Bibliography

Kasik, Reet 1996. Eesti keele sõnatuletus. Tartu Ülikooli kirjastus.


Kasik, Reet 2001. Analytic causatives in Estonian. Estonian: typological studies V (Mati Erelt, ed.).
Publications of the department of Estonian of the University of Tartu.

75
Päiviö Piia
University of Toronto

Re-Thinking Synonymy – Finnish Terminative Particles asti and saakka as


Perfect Synonyms

The term terminative refers to a case marking typical in some Finno-Ugric languages, such
as Estonian, Hungarian and the Permic languages. Terminative is generally used to designate a
morphological case that indicates a border or a limit of motion or other kind of relation. Finnish
does not have a terminative case; instead terminativity is generally expressed with the local cases
and so-called terminative particles, such as asti and saakka (‘until, all the way to, since, all the way
from, as far as’). These two particles can be seen as very close synonyms (Kangasniemi [1997: 42]
calls them perfect synonyms). If two words have same meaning, use and function, if they can be
used in same connotative, affective environments, and if their quantitative appearance in different
Finnish language corpora is identical – they are almost perfect synonyms.
In my presentation I will describe the use and distribution of asti and saakka from the oldest
written (Agricola’s) corpora to the contemporary writers of Finnish literature today (Päiviö 2007). I
will discuss the reasons, why the meaning and use of asti and saakka has remained the same, and
why for example one terminative particle has not developed new connotations and extensions to
their meaning and use. Especially, I am interested in the question, why one terminative particle is
not enough for the Finns, and what do the speakers of Finnish think about asti and saakka
introspectively?
I will base my analysis on Geeraerts (1988) article Where does Prototypicality come from?
Especially on the notion of avoiding monotony (Geeraerts 1988: 226). Additionally, I will describe
the history and development of asti and saakka and aim to trace the grammaticalization path of the
semantic meaning of these particles. The presentation offers a diachronic viewpoint to the assumed
development of the terminative particles into their present functions.

Bibliography

Geeraerts Dirk 1988: Where Does Prototypicality Come From? – Brygida Rudzka-Ostyn (ed.)
p.207-229.
Kangasniemi Heikki 1997: Sana, merkitys ja maailma. Katsaus leksikaalisen semantiikan
perusteisiin.
Oy Finn Lecture Ab, Helsinki.
Päiviö Pia 2007: Suomen kielen asti ja saakka. Terminatiivisten partikkelien synonymia, merkitys,
käyttöja kehitys sekä asema kieliopissa. TYSJYKLJ 75, Turku. (The asti and saakka of
Finnish. The Synonymy, Meaning, Use, Development and categorization of the Finnish
terminative Particles).
Rudzka-Ostyn Brygida (ed.) 1988: Topics in Cognitive Linguistics. John Benjamins, Amsterdam.

76
Rakhilina Ekaterina and Tribushinina Elena
Russian Academy of Sciences
University of Antwerp

On constructional synonymy: disentangling the Russian comparative


constructions

Cross-linguistically, it is common to express comparison by means of a similative


construction (Haspelmath and Buchholz 1998). Interestingly, Russian possesses not only a
similative construction (e.g. siât’ kak zoloto ‘shine like gold’), but also a seemingly synonymous
construction where the standard of comparison is marked by the instrumental case (e.g. siât’ zolo-
tom ‘shine gold-INS’). We hypothesized that the two constructions which are syntactically distinct
must also be semantically distinct. To test this hypothesis, we investigated the distribution of the
similative construction and the instrumental-of-comparison construction in the Russian National
Corpus (ca. 140 million words).
The results indicate that the prototypical core of the instrumental-of-comparison construction
are descriptions of shape as in letet’ klinom ‘fly in V-shape’ (lit. fly wedge-INS). Since shape is a
major visual cue to categorization (Friedrich 1970), the role of the standard of comparison in the
instrumental construction may be compared to that of numeral classifiers. Such cases do not allow a
similative in Russian. Likewise, case-free languages like English and Dutch use derivational
morphology (compounds, denominal verbs) rather than similatives for this kind of shape
descriptions as in pruimmondje trekken ‘purse one’s lips’ (lit. pull a plum-mouth).
Realizations of the instrumental-of-comparison construction beyond the domain of shape (e.g.
letet’ streloj ‘fly arrow-INS’; vyt’ volkom ‘howl wolf-INS’) denote a comparison between two
distinct entities. In such cases, the instrumental case is seemingly interchangeable with the
similative construction (cf. letet’ kak strela ‘fly like an arrow’; vyt’ kak volk ‘howl like a wolf’).
However, upon closer scrutiny it appears that the comparative instrumentals inherit several crucial
properties from the more prototypical shape-related instrumental, which distinguish them from the
corresponding similatives. One important property of the instrumental construction is monotonicity.
Shape-related instances of the instrumental construction are intrinsically monotonous – they simply
describe objects in terms of shape. Similarly, the se-mantics of the instrumental-of-comparison
beyond the domain of shape is limited to only one parameter, such as sound, manner of looking or
motion. In contrast, the similative construction often denotes comparison along several dimensions
(e.g. On vël sebâ kak podlec ‘He behaved like a scoundrel’ vs. *On vël sebâ podlecom ‘He behaved
scoundrel-INS’).
Another crucial property of the instrumental-of-comparison construction is visualness: the
comparee and the standard must be directly observable to the human eye. This explains why we
cannot use the instrumental case to express comparisons in situations which are not visually
observable as in *greet pečkoj (gives.out.warmth stove-INS) and *plavaet ryboj (swims fish-INS).
The similative must be used in such cases.
Finally, this paper will focus on the relation between the instrumental-of-comparison and
other constructions constituting an instrumental network (cf. Goldberg 1995, 2006; Kay and
Fillmore 1999) and argue that the instrumental-of-comparison is semantically connected with the
more prototypical instrumental of main predication (e.g. byl učitelem ‘was teacher-INS’) and the
instrumental of additional characterization (e.g. rabotal učitelem ‘worked teacher-INS’). We will
suggest that less prototypical predicates for expressing identity require the attributed property to be
monotonous and observable to accommodate the identity relation.

77
Bibliography

Friedrich, Paul. 1970. Shape in grammar. Language 46(2): 379–407.


Goldberg, Adele. 1995. Constructions: A Construction Grammar approach to argument structure.
Chicago: University of Chicago Press.
Goldberg, Adele. 2006. Constructions at work: The nature of generalization in language. Oxford:
Oxford University Press.
Haspelmath, Martin & Oda Buchholz. 1998. Equative and similative constructions in the languages
of Europe. In Johan van der Auwera (ed.), Adverbial constructions in the languages of
Europe, 277–334. Berlin & New York: Mouton de Gruyter.
Kay, Paul & Charles J. Fillmore. 1999. Grammatical constructions and linguistic generalizations:
The What’s X doing Y? construction. Language 75. 1–33.

78
Raukko Jarno
University of Helsinki

Evaluating and analyzing semantic similarity and sameness in studies of


polysemy and synonymy

When studying synonymy, the linguistic units whose semantic values are compared are non-
identical forms. These items (in boldface) may be given in context or in isolation:

(1a) meaning
(1b) semantics
(2a) What is the meaning of study?
(3b) What is the semantics of study?
(3a) I am interested in the meaning of synonymy.
(3b) I am interested in the semantics of synonymy.

In the case of polysemy, the object of comparison is two or more instances of one (identical)
form, often by default given in a context, or at least with contexts to be imagined:

(4a) Oh, lawyers, that's just semantics!


(4b) Oh, linguists, don't we all love semantics?
(5) (Instruction:) Can you think of two different meanings for the word semantics?

The medium of comparison is a cline from semantic sameness to semantic difference. If it is


hard to find two instances where the meaning of a form is not the same, the lexical item is not
polysemous. In normal cases of polysemy, different instances and uses of the lexical item have
sometimes different, sometimes similar meanings, and it should be the job of the semanticist to
figure out ”which distinctions matter the most”; what the different meanings of the polysemous
word (e.g., the word semantics) are:

(6a) semantics, 'the linguistic study of meaning'


(6a') in many school linguistics, an area clearly separate from syntax
(6a'') in Cognitive Linguistics, an area intertwined with syntax, pragmatics, etc.
(6b) semantics, 'the meaning content of a linguistic unit'
(6c) semantics, pejoratively used to refer to unnecessary or picky exploitation of the
nuances of the meanings of words

If the semantic difference between two uses of an identical-looking form is very big – i.e.,
there is great semantic distance – one reason may be that the case is of homonymy, not polysemy.
Another reason may be that a very polysemous word has evolved into so many directions that two
extremes in its semantic extension do not have anything in common.
In other words, the study of polysemy needs to take into account the concept of semantic
similarity in order to investigate semantic difference. In so doing, semanticists are on similar
territory when studying synonymy and polysemy. There are some psycholinguistic methods that
can be used for the purpose of covering the points of similarity and difference. One is similarity
rating, as employed on polysemy in example (7):

(7) On a scale from 0 to 5, how similar or different is the meaning of semantics in the
following examples (0 = the same, 5 = very different)?

79
Semantics studies meaning. 012345 Every word has semantics.
Every gesture has semantics. 012345 Every word has semantics.

The same method can be used for synonymy:

(8) On a scale from 0 to 5, how similar or different is the meaning of the words in
boldface in the following examples (0 = the same, 5 = very different)?

Every gesture has semantics. 012345 Every gesture has meaning.

Another method is the sorting test. Informants are given ready-made instances of one word in
different contexts, and they are asked to sort these instances into different categories according to
semantic similarity and semantic difference. This method is more readily available for studies of
polysemy, because it is easy to work on several instances of one word, but in the case of synonymy,
the items compared should include several related words in the same context, possibly
concentrating on different cases of synonymy in one study, which might distract the focus of
attention in the informant.
When using these methods in studies of polysemy, one has to evaluate their reliability and
compatibility. The concept of semantic similirity is operationalized in various ways, and in so
doing, the concept has to be evaluated more thoroughly.
One question is whether informant behavior in such experimental settings varies according to
the object of study. Do different forms suggest to us that there, by default, should be difference in
meaning? Do similar forms encourage us to presuppose that there should be similarity or at least
relatedness in meaning?
My talk will thus address the issue of semantic similarity (and sameness) from the two angles,
synonymy and polysemy, and aim at finding out whether the concept has different nuances in the
two cases by necessity.

80
Ringbom Håkan
Åbo Akademi University

Cross-linguistic similarity and the foreign language learner

Linguists have generally been more interested in linguistic differences than in similarities. But
in order to talk about difference in a meaningful way we must first consider similarity. For both
concepts a comparison is always the basis. But if no similarity at any level can be found between
two concepts, a comparison will be meaningless. It is futile to try to compare a stone with an ice
hockey match: there simply is no common ground for making such a comparison. In order to
establish meaningful differences there must be an underlying similarity. Similarity is basic,
whereas difference is secondary.
Similarities can be either cross-linguistic or intra-linguistic. I shall confine myself to cross-
linguistic similarities, but if we consider the language learner, the two go back to the same common
strategy: the learner trying to make use of prior knowledge. Learners always try to relate new
information to existing knowledge, and similarity facilitates the establishment of such relations.

For the FL learner we can distinguish three different cross-linguistic similarity relations:

(1) A similarity relation where the learner can establish a one-to-one relationship in the
target language with another unit, usually in the L1. Across related languages there will be cognates
facilitating both comprehension and learning, especially at the early stages, even though some of
them may later turn out to be false, or partially false cognates. Grammatical structures across
Germanic languages are largely similar, with the same basic categories.

(2) A difference relation where the learner can perceive both similarity and difference.
Native speakers of English learning a Romance language will encounter differences and
similarities in varying proportions.

(3) A zero relation where learners can relate few or no aspects of the target language to prior
knowledge. The zero relation does not mean that they cannot find anything at all that is relevant to
the L1 as the learning progresses. There are, after all, some linguistic universals common to all
languages. But the level of abstraction here is so high that an average language learner cannot easily
notice features that a totally different target language has in common with L1. A learner who knows
only Indo-European languages and starts learning Chinese will find it difficult to relate anything to
his prior knowledge.

Cross-linguistic similarities can be actual, perceived or assumed.

Actual similarity is a concept which at least in theory could be objectively defined by


linguistic terms. Perceived and assumed similarities, on the other hand, are psycholinguistic and
fuzzy by nature. Learners first perceive similarities to their L1. This can be fairly easy and accurate
if the L1 is related to the target language. When similarities cannot be perceived they are merely
assumed, which often leads to errors.
The different ways in which the different types of similarities affect comprehension and
production and the different language areas will be outlined. There is a lot of data from
investigations comparing Finnish and Finland-Swedish learners of English over more than 30 years.

81
Robert Stéphane
CNRS-LLACAN

Synonymy as a usefull metalinguistic pipe dream

The aim of this paper is to show (1) that strict synonymy (i.e semantic identity between two
different forms) does not exist within languages and only occasionnally across languages, but (2)
that, provided that one considers this limit, synonymy still can be a usefull tool for linguistic
analysis in two ways: first, it forces linguists to explain precisely in which way two terms can be
synonyms while differing (or conversely), and therefore to refine the description of meaning;
second, it reveals a fundamental property of language allowing speakers to construe the same
situation in alternate ways (Langacker 1991a) and, therefore, to build up equivalences within and
across languages.

In order to support these claims, I will present three components of meaning by which two
lexemes sharing a common referential value can differ (Robert 2008) and therefore are not full
synonyms:
(1) different polysemic networks
(2) different connotations
(3) different referential paths

(1) illustrates the fact that synonymy is most of the time restricted to local synonymy; it can
be examplified by the interlinguistic contrast between colour adjectives used as nouns : in English
the word greens can refer to village commons, leafy vegetables or members of a political party but
the French equivalent of greens (les verts), can only refer to members of a political party or to a set
game pawns.
(2) is of particular importance because connotations condition the context of use as well as the
pragmatic effect produced by the usage of a term; e.g. father to designate a priest signifies that the
speaker is a practicing catholic, contrary to using priest or clergyman; using the heat for
“policeman” signals belonging to a certain age group and general ideology including a pejorative
meaning in the word.
(3) is is the source of important crosslinguistic variation : languages usually choose one of the
referent’s properties to designate it, for example a physical or functional characteristic, and they
usually differ in their strategies to gain access to reference; thus English designates a “used car” not
by the fact that it is something one buys under favorable financial conditions, as in French (une
voiture d’occasion), but rather by the fact that it was previously owned (or previously owned by
only one other person as in a second-hand car).
Alternate ways to gain access to reference can be found also inside languages: for instance
pork and pig meat are two synonyms in so far as they can designate the same reference but they
differ in their referential paths. This holds true also for syntactic constructions as illustrated by the
comparison between the ditransitive construction (Bill sent Joyce a walrus) and the to construction
(Bill sent a walrus to Joyce) which accord relative salience to different facets of the construed scene
(Langacker 1991b).

Finally I will show that the different referential paths of synonyms often trigger different
constraints in their context of use. This can be illustrate by the differences between these two
syntactic constructions as well as with the two locative prepositions “ in, inside ” of Tupuri
(Ruelland 1998):

82
nen “ in, inside ” < “ eye ” → compact domain : * in a hole
bil “ in, inside ” < “ belly ” → hollow interior : * in the forest

Bibliography

Langacker, Ronald W. 1991a. Foundations of cognitive grammar (vol.2). Stanford: Stanford


University Press.
Langacker, Ronald W. Cognitive Grammar, in Linguistic Theory and Grammatical Description, F.
Droste et J. Joseph (eds), Amsterdam / Philadelphia, John Benjamins, 1991 : 275-306.
Robert, Stéphane, 2008, Words and their meanings: principles of variation and stabilization, in
Martine Vanhove (ed), From polysemy to semantic change: towards a typology of
lexical semantic associations, Studies in Language Companion Series 106. Amsterdam :
John Benjamins, 55-92.
Ruelland, Suzanne, 1998. Je pense et je parle comme je suis: le corps, le monde et la parole
entupuri. Faits de langues 11-12 : 335-58.

83
Roche Christophe and Calberg-Challot Marie
University of Savoie
Onomia - Savoie Technoloac

Synonymy in Terminology: the Contribution of Ontoterminology

Defined as a “set of designations [a designation is a “representation of a concept by a sign


which denotes it”] belonging to one special language” [ISO 1087-1], the main goal of terminology
is to eliminate ambiguity from technical languages by means of standardization.
In order to achieve such an objective, the General Theory of Terminology (GTT) postulates
several principles. One of them is bi-univocity. It means that a term denotes only one concept
(monosemy) and a concept is denoted by only one term (mononymy). Other principles are the
priority of concept over designation (term) and the universality of concept independent of the
diversity of languages.
In theory, since terminology has to be normalized, there is no synonymy in terminology (from
the GTT point of view). Nevertheless, in practice, as a term is used in texts as a word, a term owns a
sense (signified) which must not be confused with a concept – it is important to bear in mind that
the lexical structure extracted from corpora does not match the conceptual one directly built by
experts in a formal language (it means that subsumption is not hypernymy). Terms in usage are
bound into linguistic networks for different relationships, e.g. synonymy, metonymy, etc.
The aim of this article is to present the contribution of ontoterminology for understanding
synonymy in terminology. An ontoterminology is a terminology whose conceptual system is a
formal ontology – an ontology, from the knowledge engineering point of view, is a specification of
a conceptualization; it means a shared description of concepts and their relationships about a
domain expressed in a computer readable language. Ontoterminology separates and links the two
different linguistic and conceptual systems of which terminology is composed. Concepts and terms
in ontoterminology exist in their own right and definitions written in natural language (understood
as linguistic explanations) are separated from definitions written in formal language (understood as
formal specifications of concept). The conceptual level is language-independent and bi-univocity is
not mandatory – polysemy is available and only normalized terms are monosemic. It is then
possible to consider for synonymy the two dimensions, semantic and stylistic, along which the
meaning of terms can vary.
In ontoterminology, two terms are said to be synonymous if they denote the same concept –
difference in form generally corresponds to difference in usage (stylistic, sociolinguistic features).
The total synonymy, i.e. same denotation (concept) and same connotation (usage), is not so rare in a
technical domain. Ontoterminology allows to distinguish different kinds of plesionymy (structural,
functional) according to the different types of relationships between the denoted concepts (is-a,
part-of and function) and in relation to rhetorical figures (ellipsis, metonymy, meronymy,
synecdoche). For example “wheel” is a structural plesionym of “turbine” since <wheel> is a part of
<turbine>.
The article will emphasize the distinction between the linguistic and the conceptual
dimensions of terminology (a term is not a lexicalized concept and a designation is a not a
denomination). The paradigm of ontoterm (association of a concept and a term) will be introduced.
The importance of formal ontology and the role of the conceptual relationships is-a, part-of and
function for plesionymy will be detailed.

84
Bibliography

Budin, G. 2001: A critical evaluation of the state-of-the-art of Terminology Theory. ITTF


Journal,12. Vienna: TermNet.
Cabré, T. 2003: Theories in terminology. In Terminology 9(2): 163-199.
Edmonds, P., Hirst, G. 2002: Near-synonymy and lexical choice. Computational Linguistics
archive, 28(2): 105 - 144
Felber, H. 1984: Terminology Manual, Unesco (United Nations Educational Scientific and Cultural
Organization) – Infoterm (International Information Centre for Terminology).
Hamon, T., Grabar, N. 2009: Exploring graph structure for detection of reliability zones within
synonym resources: experiment with the gene ontology. Human Language Technology,
Proceedings of the Workshop on BioNLP, Boulder, Colorado, June 2009: 89–96.
Hirst, G. 1995: Near-synonymy and the structure of lexical knowledge. In AAAI Symposium on
Representation and Acquisition of Lexical Knowledge: Polysemy, Ambiguity, and
Generativity, Stanford: 51–56,
ISO 1087-1:2000. Terminology work-Vocabulary-Part1: Theory and application. International
Organization for Standardization.
ISO 704:2000. Terminology work - Principles and methods. International Organization for
Standardization.
Pavel, S. & Nolet, D. 2001. Handbook of Terminology. Minister of Public Works and Government
Services Canada 2001, Catalogue No. S53-28/2001.
Roche, C., Calberg-Challot, M., Damas, L., Rouard, P. 2009. Ontoterminology: A new paradigm
for terminology. International Conference on Knowledge Engineering and Ontology
Development, 5-8 October, Madeira (Portugal).
Wright, S.E., Budin, G. 1997: Handbook of Terminology Management, volume 1 and 2, John
Benjamins Publishing Company.

85
Samedova Nezrin
Azerbaijan University of Languages

On Two Russian Constructions: What Else If Not Synonyms?

1. The semantic relationships of the Russian construction nachat’+INF and perfective


construction stat’+INF1 have been attracting linguists’ attention for a long time. It has been pointed
out that they substitute each other. At the same time, researchers do not find their meanings
completely identical and attempt to describe their semantic singularity. However, there are not
unanimously accepted semantic descriptions of these constructions.
2. I have investigated the question using the method of system analysis [Ломтев 1976].
2.1. Both constructions have the meaning of the beginning (initiality) and are translated into
English as to begin+INF. Both are widely used and the areas of their usage fully coincide. The main
components of both are perfective, whereas only imperfectives are permitted as their non-main
components. It is clear, however, that the semantic identity of these constructions cannot be
absolute.
Indeed, stat’+INF is used much more frequently than nachat’+inf, cf. Russian National
Corpus. Hence, according to the Kruszewski – Kuryłowicz rule [Березин 1998: 15] we can state
that stat’+INF is semantically simpler than nachat’+INF.
2.2. The comparison of nachat’ has with its imperfective correlate nachinat’ enables to assert
that the construction nachat’+INF has the following semantic structure 2:
1) the non-aspectual meaning of a non-homogeneous process (initiality)
– and correspondingly –
2) the aspectual meaning of the final moment (perfectivity);
3) the non-aspectual meaning of the process inherent in an infinitive.
If the metaphor of line is used for interpreting the concept of a process and the metaphor of
point 3 is used to interpret the concept of perfectivity, we can illustrate the meaning of nachat’+INF
this way:

initiality (non-aspectual) a process


—————————————•—————————————
perfectivity

2.3. As regards stat’+INF, we face a paradox.


On the one hand, the meaning of its components contains three elements in total: two of them
belong to the verb stat’ (the non-aspectual meaning of a process and perfectivity) and one belongs
to the infinitive (the non-aspectual meaning of a process). On the other hand, the meaning of
stat’+INF contains fewer than three components, for it is simpler than the meaning of
nachat’+INF.
G.S. Samedov has solved the paradox. The thing is that two meanings of a process merge,
thus stat’+INF has the meaning of a syncretic process. The merger is possible because the nature of
the initiality possessed by stat’+INF is different from that of characterizing nachat’+INF. It is like
a point. In other words, the perfectivity attributing to the verb stat’ is the meaning of the initial
moment, or punctual initiality. That is why it does not prevent the meanings of a process from
amalgamating:

1
Not to be confused with the homonymous imperfective construction stat’+INF.
2
We ignore irrelevant details.
3
Both metaphors possess a specific cognitive content.

86
initiality (aspectual),
or perfectivity

•——————————————————————

a syncretic process

Thus, the meaning of the perfective construction stat’+INF indeed contains two components.
3. Besides being a special interest for languages that have the category of aspect, the analysis
enables to make the following fundamental conclusion. Linguistics does need the concept of
synonymy.
As to the nature of the phenomenon, I believe the analyzed case has shed new light on it.

Bibliography

Березин, Ф.М. 1998: Н.В. Крушевский – провозвестник лингвистики ХХ в. Н.В. Крушев-


ский. Избранные работы по языкознанию. Moscow: Наследие. 4-24.
Ломтев, Т.П. 1976: Общее и русское языкознание. Moscow: Наука.

87
Schmeiser Benjamin
Illinois State University

Synonymy in Contemporary United States Spanish

A topic of current interest in Spanish linguistics is that of lexical (i.e. ‘traditional’) synonymy
in diachronic terms. Recent studies (Stale, 2009; Horcas Villareal, 2009) have concentrated on the
occurrence of synonymy in diachronic terms. As Spanish evolved from Vulgar Latin, synonymy
resulted due to word transmission. That is, patrimonial words (voces populares), those words that
underwent the predicted phonological processes, also maintained their learned word (cultismo) or
semi-learned word (semicultismo) counterpart, as in (1).
The aforementioned studies have greatly increased our diachronic understanding of lexical
synonymy in Spanish. That said, however, it is surprising to note the lack of research treating
synchronic analysis of functional synonymy. For instance, Contemporary United States Spanish
(henceforth, CUSP) is known for its abundance of synonymous lexical entries, as in (2). Unlike its
diachronic counterpart, however, synonymy does not occur as a result of word transmission. That is,
a peculiar trend emerges from the list above in (2) in that the words both on the left and the right
already exist in Standard Spanish and have different meanings. For example, both la carpeta means
‘the folder’ and la alfrombra means ‘the carpet’. However, in the case of CUSP, semantic extension
as a result of contact with English has occurred, thus forming a doublet (doblete) in that la carpeta
means both ‘the folder’ and ‘the carpet’. This is particularly relevant in that synonymy occurs in
this variety from words generally that already exist in the language. For instance, vacunar exists in
Standard Spanish as ‘to vaccinate’, with pasar la aspiradora meaning ‘to vacuum,’ yet in CUSP
vacunar is used in its more functional meaning and pasar la aspiradora is understood, though
generally not produced.
I conclude my treatment of the relationship between lexical and functional synonymy by
suggesting that, although the transmissions were quite different, a similar pattern emerges; in
diachronic terms, the patrimonial word (voz popular) is generally the more frequently used word,
with the (semi)learned word ((semi)cultismo) less frequently used. In synchronic terms, the word
that undergoes semantic extension as a result of contact with English is generally the more
frequently used word, with the Standard Spanish word less frequently used.
In short, the current study is novel in that it is the first known study to consider the
relationship between lexical and functional synonymy. In addition, it greatly adds to the field by
exploring a linguistic process and a variety of Spanish that have both been inexplicably
understudied. In the study, I illustrate how synonymy in Spanish has had two distinct patterns,
namely i) during the formation of the language via word transmissions from Latin and ii) in a
modern variety of Spanish that is in contact with English. Additionally, I offer a linguistic analysis
that considers both the differences (e.g. word transmission vs. semantic extension) and the
similarities (e.g. a trend for one of the lexical entries to be much more common) of both types of
synonymy.

(1) patrimonial: hostigar (semi)learned: fustigar ‘to pester’


injertar insertar ‘to graft’
lindar limitar ‘to restrict’
(2) la luz el semáforo ‘the traffic light’
la carpeta la alfrombra ‘the carpet’
vacunar pasar la aspiradora ‘to vacuum’
aplicar solicitar ‘to apply’

88
Bibliography

Calvi, Maria Vittoria and Emma Martinell (1997). Los dobletes léxicos en la enseñanza del español
a extranjeros. ASELE Actas VIII, 227-239.
Horcas Villareal, José Mario (2009). El español: dobletes, cultismos, y neologismos.
Contribuciones a las Ciencias Sociales, 1-5.
Penny, Ralph (2004). Variation and Change in Spanish. Cambridge University Press: Cambridge.
Stala, Ewa (2009). Dobletes etimológicos en español - su origen y evolución semantica:
observaciones puntuales. Studia linguistica, 126, 113-127.

89
Sirola-Belliard Maija
University of Tampere

The Finnish comitative case -ine and the postpositional kanssa construction –
how synonymous are they?

The paper examines the relation between the Finnish comitative case -ine and the construction
with the postposition kanssa 'with' that have been assumed to be synonymous. The core meaning of
the comitative is Accompaniment, although cross-linguistically the same form can also be used for
encoding Instrument, Possession or Inclusion, for example. Accompaniment can be expressed by
adpositions, case affixes and serial constructions, among other means. (Stolz et al. 2009: 602f.) In
Finnish, the principal means are the number-insensitive comitative case affix -ine which is
accompanied by a possessive suffix when attached to a noun, as in example (1), and several
postpositions governing the genitive case, the main variant being kanssa, as in example (2).
In recent literature it has been claimed that the functional domains of the comitative case and
the postposition kanssa overlap (Stolz et al. 2006: 61, Karlsson 1982: 132; partly also Haarala et al.
1990–1994, Hakulinen et al. 2004: 942f., 1211f.) and that the comitative case is being replaced by
the postposition kanssa (Stolz et al. 2005: 214, 2006: 61; cf. also Nau 1995: 133), which implies the
assumption of synonymy between the two constructions. In my presentation, I will show that these
claims are questionable.
It is clear that the two constructions have common functions, such as the prototypic function
of Accompaniment. For the most part, however, newspaper corpus material (Pajunen 2003) shows
that their functional domains are different: the frequencies of the common functions are notably
different and, in addition, both constructions have some distinct functions that cannot be expressed
by the other construction. Essentially, kanssa concentrates on expressing Accompaniment
extensively while the case is used in more diverse functions (cf. Sirola 2008). In my paper, I will
demonstrate this on the basis of corpus material.
The differences in the usage of the two constructions are directly based on their meanings.
First, the construction with kanssa combines two symmetric, independent participants, whereas the
comitative case represents an asymmetric relationship in which the companion (marked with the
case affix) is subordinate to the accompanee. Consequently, it is impossible to encode for example a
reciprocal action with the inflectional comitative while it is commonly expressed by kanssa.
Second, since the comitative case marker is followed by a possessive suffix that refers (in most
cases) to the accompanee, the participants are semantically bound to each other; the participants of
the structure with kanssa do not have such a semantic limitation. On the one hand, this favors the
postposition kanssa, as it can be used more widely for expressing Accompaniment, for example
with proper nouns. On the other hand, the bond between the participants allows the comitative case
to express diverse kinds of meronymic and hyponymic relations which cannot be expressed by the
postposition kanssa.

Examples (stem from the HS2000 corpus, Pajunen 2003):

(1) Anne Nordlund laps-ine-en on palannut kuukausi sitten Suome-en.


Anne Nordlund child-COM-POSS.3 has returned month ago Finland-ILL
'Anne Nordlund has returned to Finland with her child(ren) a month ago.'
(HS 1 578 563)

(2) Äiti oli lapse-nsa kanssa polkupyör-i-llä matka-lla uimaranna-lle.


mother was child-GEN+POSS.3 with bicycle-PL-ADE way-ADE beach-ALL
'Mother was on her way to the beach with her child by bicycles.'
(HS 443 214)

90
Bibliography

Haarala, R.; Lehtinen, M.; Grönros, E.-R.; Kolehmainen, T. and Nissinen, I. (eds) 1990–1994.
Suomen kielen perussanakirja. Kotimaisten kielten tutkimuskeskuksen julkaisuja 55.
Helsinki: Kotimaisten kielten tutkimuskeskus.
Hakulinen, A.; Vilkuna, M.; Korhonen, R.; Koivisto, V.; Heinonen, T.-R. and Alho, I. (eds) 2004.
Iso suomen kielioppi. Suomalaisen Kirjallisuuden Seuran toimituksia 950. Helsinki:
SKS.
Karlsson, F. 1982. Suomen peruskielioppi. Suomalaisen Kirjallisuuden Seuran toimituksia 378.
Helsinki: SKS.
Nau, N. 1995. Möglichkeiten und Mechanismen kontaktbewegten Sprachwandels unter besonderer
Berücksichtigung des Finnischen. Edition Linguistik 08. München – Newcastle: Lincom
Europa.
Pajunen, A. 2003. HS2000. 31 million word FDG-analysed Context Mill database of the newspaper
Helsingin Sanomat, volumes 2000 and 2001.
Sirola, M. 2008. Komitatiivi nykysuomessa: sijan typologista ja areaalista taustaa sekä sen
ilmaisemat merkitykset Helsingin Sanomien korpuksessa. Master's thesis. University of
Tampere.
Stolz, T.; Stroh, C. and Urdze, A. 2005. Comitatives and Instrumentals, The World Atlas of
Language Structures, M. Haspelmath, M.S. Dryer, D. Gil, B. Comrie (eds), Oxford:
Oxford University, 214–217.
Stolz, T.; Stroh, C. and Urdze, A. 2006. On comitatives and related categories: a typological study
with special focus on the languages of Europe. Berlin – New York: Mouton de Gruyter.
Stolz, T.; Stroh, C. and Urdze, A. 2009. Varieties of comitative, The Oxford Handbook of Case, A.
Malchukov and A. Spencer (eds), Oxford – New York: Oxford University Press, 601–
608.

91
Soares da Silva Augusto
Catholic University of Portugal

Competitions of synonyms through time:


Conceptual and social salience factors and their interrelations

This paper takes as a starting point three different hypotheses. First, the study of the
diachronic development of synonymous forms reveals essential aspects about the nature and
motivations of synonymy. Second, the emergence and competition of synonymous forms can either
result from conceptual salience factors, namely prototypicality (semasiological salience) and
entrenchment (onomasiological salience), or from social salience factors, i.e. sociolinguistic,
stylistic or pragmatic prevalence, or even from interaction of both salience factors. Crucially,
prototype-theoretical features of the concepts involved can determine the necessary differences
between synonyms within and across lectal varieties or even across languages; and the lectal
features of the items involved can not only determine the occurrence of synonyms across lectal
varieties but also motivate the differences of prototypical structure between synonyms. The third
hypotheses is that competitions of synonyms shed light about the processes of language variation
and change, including convergence and divergence processes between lectal varieties and the
processes of linguistic stratification and standardization.
These hypotheses will be tested with two different corpus-based case studies in lexical
synonymics of Portuguese. The first case study is about the semantic development of the verb
deixar ‘to leave, to let’ from Old to Modern Portuguese and its most competitive synonyms, namely
abandonar ‘to abandon’ (a Gallicism) and permitir ‘to allow, to permit’ (a juridical Latinism). The
late entry of abandonar and permitir into Portuguese (in the fifteenth century only) led to a situation
of full conceptual and distributional synonymy with the two prototypical uses of deixar at the time.
However, this situation rapidly caused a semantic dissimilation which took the shape of a prototype
reorganization, mainly in the semasiological structure de deixar: the conceptual center of the
subgroup of senses of deixar with verbal complement shifted from the active sense of ‘to allow’ to
the passive sense of ‘not to intervene’ and the two prototypical centers of deixar climbed up to
hierarchically more abstract levels. At the same time, nondenotational meaning differences, namely
stylistic, emotional and interactional have facilitated these prototype reorganizations. Additionally,
we will see that Romance cognate verbs, namely the Portuguese deixar, Spanish dejar, French
laisser, Italian lasciare and Romanian a lăsa are not totally semantically equivalent, as they also
have differences in prototype organization.
The second case study includes the development of four dozens of sets of denotational
synonymous nouns selected from the lexical fields of football and clothing in European and
Brazilian Portuguese in the last 60 years. The aim is to examine the impact of item-related features
(i.e endo-/exogenousness, foreign influence or loanwords, archaism, neologism) and concept-
related features (i.e. prototypicality, vagueness, recent origin, semantic field) in the production of
denotational synonyms within and across the two national varieties of Portuguese. Internal
uniformity measures enable us to measure the lexical uniformity/diversity of a certain concept in a
language variety: the internal uniformity value will decrease the more synonymous terms there are
competing to denote the same concept, and the more dominant some of these terms become. These
measures allow us to calculate increases/decreases of lexical homogeneity/heterogeneity in both
national varieties of Portuguese and lexical standardization process as well, and they also permit to
compare the degree of change. External uniformity measures enable us to know if the two national
varieties of Portuguese have been going through a process of lexical convergence or divergence.
The denotational synonyms studied show that the Brazilian variety has changed more than the
European variety and that both varieties diverge from each other in the vocabulary of clothing.

92
The two diachronic studies stress the following characteristics of synonymy: the role of
prototypicality in the differentiation of synonyms, the sufficient semantic similarity and the
necessary semantic differences between synonyms, the superfluity and contingency of total
synonymy, the influence and correlations of concept-related features and item-related features in the
emergence and competition of synonyms within and across lectal varieties, and the interaction of
conceptual and social factors in the occurrence of synonyms.

93
Sutrop Urmas
Institute of Estonian Language, Tallin
University of Tartu

Synonymy of fuzzy and scalar concepts and terms

Studying vocabulary one can see that different lexical domains behave differently. For
example, colour terms correspond to colour concepts that are fuzzy “clouds” in the colour space,
but temperature terms form scalar concepts. In both cases one can find synonyms, basic and non-
basic terms. Even very simple categories such as white and black have synonyms – whitish, snow
white, warm white, charcoal black, deep black, etc.
In an empirical study of Estonian sense perception terms 80 subjects were interviewed both
for colour and temperature terms. In colour naming task each subject 65 standard Color-aid colour
tiles were shown and asked to name them. As a result, 5,197 names were given to the tiles at all.
Among these there were 638 different terms. As a mean 20.1 synonyms were given for each tile.
The minimum number of synonyms (9) got tile Y (yellow), and the maximum number of synonyms
(41) got ORO S3 (orange-red-orange shadow 3). It is clear that colour tiles with focal colour have
fewer synonyms and non-focal colours have more synonyms. The distribution of synonyms for each
tile corresponds to Zipf curve. Basic colour terms tend to be most frequent dominant synonyms and
non-basic recessive synonyms occur with law frequency.
In the temperature term ranking task 80 subjects were asked to build up a temperature scale
using temperature terms she or he had named in the earlier tasks. Every term in a scale got
numerical value. Psychological zero (decided arbitrarily) got the value 0. As a result a progression...
-2, -1, 0, 1, 2 ... follows. After that the subjective temperature scale was constructed over all
individual scales. In the ranking task 659 terms were used in all. Among these there were only 85
different terms in comparison with the 142 different terms listed in the earlier tasks. Every
individual temperature scale consisted of a mean 8.24 terms.
In the case of fuzzy colour concepts the synonymy was defined using standard colour tiles.
All names given to a certain colour tile were synonyms. In the case of scalar temperature terms one
can define the terms which are located “quite close” on the subjective integral temperature scale as
synonyms. So ice cold and icy, suitable and tepid, and burning and hot are synonyms in Estonian.
They do not denote the same temperature.

94
Stede Manfred
University of Potsdam

TBA

95
Wang Tong and Hirst Graeme
Department of Computer Science, University of Toronto
University of Toronto

Associating Difficulty in Near-Synonymy Choice with Types of Nuance


Using Core Vocabulary

There are arguably infinitely many dimensions along which members of a cluster of near-
synonyms can differ (Cruse, 1986), and the nuances that differentiate near-synonyms along a
dimension are often subtle and difficult even for native speakers. The diversity of near-synonym
variation types has motivated the categorization of these dimensions of these variations. DiMarco,
Hirst, & Stede (1993) proposed 38 dimensions for differentiating near-synonyms, which were
further categorized into semantic and stylistic variations. Stede (1993) focused on the latter and
further decomposed them into seven scalable sub-categories. Inkpen & Hirst (2006) organized
near-synonym variations into a hierarchical structure, combining stylistic and attitudinal variation
into one class in parallel to denotational differences.
Despite the variety of categorization methods, stylistic variation among near-synonyms is an
important dimension that has been frequently addressed. In this study, we hypothesize that the
stylistic nature of nuances correlates to the degree of difficulty in choosing between near-synonyms.
Contrasting some recent studies that focus on contextual preferences of synonyms (e.g., Arppe &
Järvikivi 2007), we elect to investigate the internal features of near-synonym nuances. Specifically,
we adopt the notion of core vocabulary to associate stylistic variation in theory with the difficulty
level of near-synonym choice in practice. Core vocabulary consists of “words that suffice to define
all of the remaining vocabulary of a language” (Lehmann 1991). It was first related to stylistic
variation among near-synonyms by Stede (1993). Carter (1987) listed ten features of core
vocabulary, among which, associationism is the “bridging” dimension between stylistic variation
and core vocabulary. It is characterized by scalable dimensions closely resembling those Stede used
for characterizing stylistic variations. Carter claimed that core vocabulary words are relatively
neutral on these scales, indicating fewer stylistic variations among them.
Notably, some of Carter’s features of CV are readily verifiable using computational linguistic
techniques. The collocability of a word, for example, can be approximated by the number of co-
occurring word types (normalized by the number of senses to eliminate the confounding factor of
polysemy); neutrality in field of discourse can be verified by a word’s distribution across different
genres in a balanced corpus. Multiple linguistic resources are combined in our study to achieve an
empirical characterization of core vocabulary.
To test our hypothesis, a near-synonym lexical choice task (Edmonds 1997) is employed to
measure difficulty levels. In this task, lexical gaps are created in sentences from a corpus by
removing members of a near-synonym cluster. The sentences are then presented to subjects whose
task is to determine from context which member of the cluster is the missing word. Experiments in
existing studies have shown great variance in the performance (and hence in level of difficulty) on
different near-synonym clusters (Edmonds 1997; Inkpen 2007). Our study shows that such variance
is correlated with differing degrees of coreness of the near-synonyms, and in turn, different types of
near-synonym variations. Counter to intuition, the seemingly subtle stylistic nuances are usually
easier for subjects to distinguish than non-stylistic differences.

96
Bibliography

Arppe, Antti & Järvikivi, Juhani. 2007: Every method counts: combining corpus-based and
experimental evidence in the study of synonymy. Corpus Linguistics and Linguistic
Theory, 3(2), 131–159.
Carter, Ronald. 1987: Vocabulary: Applied Linguistic Perspectives. Allen & Unwin.
Cruse, D. A. 1986: Lexical Semantics. Cambridge University Press.
DiMarco, Chrysanne; Hirst, Graeme; & Stede, Manfred. 1993: The semantic and stylistic
differentiation of synonyms and near-synonyms. AAAI Spring Symposium on Building
Lexicons for Machine Translation, 114–121.
Edmonds, Philip. 1997: Choosing the word most typical in context using a lexical co-occurrence
network. Proceedings of the 35th Annual Meeting of the Association for Computational
Linguistics and Eighth Conference of the European Chapter of the Association for
Computational Linguistics, 507–509.
Inkpen, Diana. 2007: A statistical model for near-synonym choice. ACM Transactions on Speech
and Language Processing, 4, 1–17.
Inkpen, Diana & Hirst, Graeme. 2006: Building and using a lexical knowledge base of near-
synonym differences. Computational Linguistics. 32, 223–262.
Lehmann, Hubert. 1991: Towards a core vocabulary for a natural language system. Proceedings of
the Fifth Conference of the European Chapter of the Association for Computational
Linguistics, 303–305.
Stede, Manfred. 1993: Lexical choice criteria in language generation. Proceedings of the Sixth
Conference of the European Chapter of the Association for Computational Linguistics,
454–459.

97
Wiemer Björn and Socka Anna
Johannes-Gutenberg-Universität
University of Gdansk

How to do contrastive semantics with propositional modifiers:


The case of hearsay adverbs

The paper will present preliminary results from two ongoing projects dealing with
evidentiality markers. Specifically, our concern is the semantic description of a couple of
propositional modifiers indicating hearsay in Polish and German (Pol. rzekomo, jakoby, podobno,
Germ. angeblich, vorgeblich, mutmaßlich) which – though considered close synonyms – often
cannot faithfully be translated into each other, nor by Eng. allegedly. These hearsay markers have
been claimed to carry epistemic overtones by which the actual speaker transmits his/her doubts into
the contents of the message referred to (P). Corpus-based research shows that these lexemes do so
to a varying extent: with some of them, e.g. podobno, epistemic overtones arise only via
conversational implicature, while for others, e.g. rzekomo, they seem to form part of their inherent
semantics and are not cancellable in most contexts (Wiemer 2006). This analysis leads to an
ordering along degrees of epistemic strength that resembles Horn-scales:

Figure 1: Increase of epistemic strength (indicated by arrow)


(1a) ‘I think that P might be not true.ʼ Pol. podobno
(1b) I think that P can be not true.ʼ jakoby
(1c) ‘I think that P is not true.ʼ rzekomo

However, Socka (2009; 2010), in her corpus-based study on Polish-German translational


equivalence, has shown that any of the aforementioned hearsay markers can become void of
epistemic overtones in specific contexts; the „negative” epistemic default can be cancelled even for
Pol. rzekomo and Germ. angeblich (see below). For a similar observation regarding Eng. allegedly
cf. Ramat/Ricca (1998: 230), who speculated that the epistemic value of the respective sentence
adverb is influenced mainly by the existence (or absence) of alternative hearsay adverbs in the
language which „share“ into the stages of a scale corresponding to Fig. 1. They assume an analogy
between merely epistemic adverbs and reportive adverbs (often carrying a negative epistemic
commitment) to rest on Horn-scales, each with a marked and an unmarked member:

Figure 2: Horn-scale based analogy between epistemic and reportive adverbs

‘probably’ : ‘allegedly’

‘possibly / perhaps’ : ‘reportedly’

subjective assessment of

[likelihood of P being true] : [likelihood of P being untrue]

marked member of opposition (arrow indicates increase of negative stance as for P being true)

98
We can show that the epistemic load of the aforementioned hearsay markers cannot be
captured by Horn-scales. Explanations based on the Q-principle (cf. Levinson 2000; Huang 2007)
must fail unless they arise from clausal implicatures. Best suited are explanations based on the M-
principle, applied in practice by Olbertz (2007) for Mexican Spanish dizque. Olbertz‘ pragmatic
analysis, in turn, does not explain (and even contradicts) facts known from the distribution of such
hearsay markers across discourse types: Germ. angeblich and Pol. rzekomo regularly lose their
otherwise strong epistemic default in news reports, i.e. in discourse for which an indication of
second-hand information is presupposed and, on this account, redundant.
Given these puzzles on the borderline between lexical semantics and pragmatics, we
formulate a proposal of how epistemic and evidential meaning components should be accounted for
in the representation of German and Polish hearsay adverbs (and particles). This case study also
allows to highlight general problems raised by units with „subjective“ meanings that modify entities
on the propositional and higher layers (in Dik et al.‘s 1990 terms): since their description and (intra-
lingual as well as cross-linguistic) comparison can be only intensional, translational equivalence is
here even less sufficient to establish synonymy than in the case of figurative extensions of
adjectives denoting physical properties or kinship terminology. We thus use our case study to
illustrate how the problem of synonymy raised in lexical typology (cf. Koptjevskaja-Tamm 2008,
Evans 2010) can be expanded to function words.

Bibliography

Dik, S.C., Hengeveld, K., Vester, E., Vet, C. (1990): The hierarchical structure of the clause and the
typology of adverbial satellites. In: Nuyts, J., Bolkestein, M., Vet, C. (eds.): Layers and
levels of representation in language theory. Amsterdam, Philadelphia: Benjamins, 25-
70.
Evans, N. (2010): Semantic typology. In: Jung Sung, J. (ed.): The Oxford Handbook of Linguistic
Typology (to appear). Oxford etc.: Oxford U.P., 504-533.
Huang, Y. (2007): Pragmatics. Oxford etc.: Oxford U.P.
Koptjevskaja-Tamm, M. (2008): Approaching lexical typology. In: Vanhove, Martine (ed.): From
Polysemy to Semantic Change (Towards a typology of lexical semantic associations).
Amsterdam, Philadelphia: Benjamins, 3-52.
Levinson, S.C. (2000): Presumptive meanings. The theory of generalized conversational
implicature. Cambridge, M.A.: MIT Press.
Olbertz, H. (2007): Dizque in Mexican Spanish: the subjectification of reportative meaning. In:
Squartini, M. (ed.): Evidentiality between lexicon and grammar. Italian Journal of
Lingustics – Rivista di linguistica 19, 151-172.
Ramat, Paolo, Ricca, Davide (1998): Sentence adverbs in the languages of Europe. In: van der
Auwera, Johan, Baoill, Dónall P.Ó. (eds.): Adverbial Constructions in the Languages of
Europe. Berlin, New York: Mouton de Gruyter, 187-275.
Socka, A. (2009): Marker der reportativen Evidentialität im Deutschen und ihre polnischen
Äquivalente. In: Akten des 43. Linguistischen Kolloquiums „Pragmantax II. Zum
aktuellen Stand der Linguistik und ihrer Teildisziplinen“, Magdeburg, 10.-13.09.2008.
Socka, A. (2010): Reportative Partikeln in kontrastiver Sicht (Polnisch – Deutsch). In: Kątny, A.,
Socka, A. (eds.): Modalität / Temporalität in kontrastiver und typologischer Sicht.
Frankfurt/M. etc.: Lang, 239-264.
Wiemer, B. (2006): Particles, parentheticals, conjunctions and prepositions as evidentiality markers
in contemporary Polish (A first exploratory study). Studies in Polish Linguistics 3, 5-67.

99
Vilkki Liisa
University of Helsinki

The comparability of similar inferential meanings across languages

Inferential expressions indicate the speaker’s inference, based on some type of direct
information source, such as observational or memorial source. Different kinds of inferential
expressions show remarkable cross-linguistic semantic variation. However, genetically related
languages or languages in contact often have inferential expressions that seem to indicate quite
similar, even (near) synonymous meanings. The purpose of this paper is to discuss the
comparability of inferential meanings in languages by means of comparative concepts (cf.
Haspelmath 2010) and the degree of similarity between language-specific inferential meanings. The
discussion is based on my typological investigation into the semantic domains of epistemic
modality and inferentiality as they are grammatically represented in the world’s languages. This
study uses descriptive data from the sample of 130 languages. In this kind of study, it is not possible
to consider in great detail contextual and stylistic differences between the compared meanings.
Nevertheless, it is possible to deal with the notions of similarity and synonymy from various
perspectives. In this presentation, I will firstly introduce the comparative concepts, distinguished for
language-specific grammatical inferential meanings, and illustrate the relationship between the
comparative concepts and meanings by several examples. Second, I will focus on the question of
how differences between semantic networks arguably affect the degree of similarity between
grammatical inferential meanings.
The central type of comparative concept, used in this study, is that of function. This notion
is defined by Haspelmath (2003) as neutral between language-specific notions of sense and use (or
meaning). According to Haspelmath, a function is distinguished if there is at least one language that
has a formal expression for that function. Due to the differences in quality and details of
information, provided by the descriptive material, the functions created in this study have to be
somewhat more coarse-grained. However, they capture the main distinctions of the variation in
meaning, and the proposed semantic description is more fine-grained than in the previous large-
scale typological studies concerning inferentiality and epistemic modality. The pure inferential
functions do not include epistemic properties, indicating degrees of the speaker’s certainty. They
have been elaborated on the basis of epistemological notions of direct sources of
knowledge/information which can provide a foundation for an inference (e.g. Audi 2003).
Inferential functions can also consist of inferential properties combined with epistemic properties.
For instance, the English modal must represents several types of functions with combined
properties. The basic purpose of functions is to identify similar meanings within languages and
across languages. Actually, they cover meanings, ranging from (near) synonymous to roughly
similar meanings. For example, the function ’inference from memory’ represents the gamut of
inferential meanings, such as ‘inference based on previous experience’ and ‘inference based on
general knowledge’. Inferential meanings of different languages are members of various types of
semantic networks. I will argue that if the notion of (near) synonymy is needed at all instead of the
notion of similarity in the comparison of inferential meanings across languages, it should be
confined to those inferential meanings that belong to highly similar semantic networks.

Bibliography

Audi, Robert 2003. Epistemology. A Contemporary Introduction to the Theory of Knowledge. New
York, London: Routledge.

100
Haspelmath, Martin 2010. Comparative Concepts and Descriptive Categories in Cross-Linguistic
Studies. To appear in Language 86.
Haspelmath, Martin 2003. The Geometry of Grammatical Meaning: Semantic Maps and Cross-
Linguistic Comparison. The New Psychology of Language, Volume 2., Michael
Tomasello (ed.), Mahwah, New Jersey: Lawrence Erlbaum Associates, 211-242.

101
Västi Katja
University of Oulu & University of Helsinki

Searching for constructional synonymy: Paraphrasing as evidence against


ellipsis

In this presentation, the concept of synonymy is used to signal essentially identical reference
to particular event types by separate clause-level constructions. By “essentially”, it is meant that the
events cover the same profiled participants (in the sense of cognitive grammar and frame semantics,
see, e.g., Fillmore 1977, 1982; Langacker 1987). More precisely, I discuss the meaning of the
Finnish ablative-initial (cf. 1) and allative-initial (cf. 2) verbless constructions that are functionally
comparable to finite clauses and occur frequently in headlines. Since they lack a finite verb, it is
challenging to indicate the event types they can refer to: traditionally, it is taken that event types are
largely specified by verbs. However, paraphrasing is seen as one solution for describing the
meaning of the constructions in question and finding synonymous finite constructions for them.

(1) Opiskelijo-i-lta monikulttuurinen vaatemallisto


student-PL-ABL multicultural[NOM.SG] apparel.collection[NOM.SG]
lit. ‘From students a multicultural apparel collection.’

(2) Varka-i-lle iso saalis ravirada-lta


thief-PL-ALL big[NOM.SG] haul[NOM.SG] trotting-track-ABL
lit. ‘To thieves a big haul from trotting-track.’

There are comparable finite constructions in Finnish, i.e., constructions with the same cases in
the same positions and a finite verb (e.g., the passive construction, the necessive zero subject
construction, and the possessive construction), but, based on an intuitive analysis, I have argued that
the ablative-initial and allative-initial verbless constructions cannot be seen as elliptic clauses. That
is, these verbless constructions denote different event types from the comparable finite
constructions; it is not the case that a verb is simply omitted. To justify the argument also
empirically, I have designed a paraphrase test for discovering how the ablative-initial and allative-
initial verbless constructions are construed. In order for the verbless constructions to be elliptic,
most of the paraphrases would have to be formally identical except for a “restored” verb:
respondents would intuitively know the “missing” verb. 163 respondents participated in the
paraphrase test concerning the allative-initial verbless constructions and 209 respondents in the one
concerning the ablative-initial verbless constructions, and most of the yielded paraphrases were
basic transitive or intransitive clauses (cf. 3) – that is, not the constructions that simply adding a
verb would have yielded. This is crucial for establishing that they are not elliptic clauses but
independent constructions (in the sense of, e.g., Goldberg 1995, 2006) that are verbless from the
very beginning.

(3) varkaa-t sa-i-vat iso-n saalii-n ravirada-lta


thief-NOM.PL get-PST-3PL big-ACC haul-ACC trotting-track-ABL
‘Thieves had a big haul from a/the trotting-track.’

Thus paraphrasing verbless constructions with finite clauses illustrates that synonymy may be
defined as covering far broader domains than individual lexical elements; it applies to coding
particular event types in different ways as well. In that case, synonymy must certainly be seen as a
relatively coarse-grained equivalence between constructions. It does not cover, for example,
102
pragmatic aspects of meaning. However, from my point of view, this is not a problem but the
desired state of affairs: in order for synonymy to be a useful concept in linguistics, it needs to be
inclusive enough.

Bibliography

Fillmore, Charles J. 1977. Topics in Lexical Semantics. Current Issues in Linguistic Theory, Roger
W. Cole (ed.), 76–138. Bloomington and London: Indiana University Press.
Fillmore, Charles J. 1982. Frame Semantics. Linguistics in the Morning Calm: Selected Papers
from SICOL-1981, The Linguistic Society of Korea (ed.). Seoul: Hanshin, 111–137.
Goldberg, Adele E. 1995. Constructions: A Construction Grammar Approach to Argument
Structure [Cognitive Theory of Language and Culture]. Chicago and London: The
University of Chicago Press.
Goldberg, Adele E. 2006. Constructions at Work: The Nature of Generalization in Language.
Oxford: Oxford University Press.
Langacker, Ronald W. 1987. Foundations of Cognitive Grammar [volume I]: Theoretical
Prerequisites. Stanford: Stanford University Press.

103
Zenner Eline, Speelman Dirk and Geeraerts Dirk
University of Leuven
University of Leuven
University of Leuven

Lexical Choice and Loanwords: the Use of Anglicisms in Dutch

Synonymy and the Success of Loanwords: In this paper we discuss the issue of synonymy
from a contact linguistics angle. Our focus is on determining which features influence the relative
preference for a loanword when a synonymous native alternative exists. This onomasiological
perspective on borrowing is developed in two separate case studies, each dealing with English
person reference nouns (e.g. designer, lover) in a Belgian Dutch and Netherlandic Dutch newspaper
corpus (consisting of over 1 billion words ranging from 1999 to 2005).
Case Study 1 – Conceptual Features: The first study tries to capture the influence of
conceptual features on the success of English person reference nouns. We define that success as the
corpus frequency of the anglicism, relative to the the total frequency of the English loan word and
its denotational synonyms:

Synonymous expressions for the Token frequencies in the


concept HOOLIGAN Dutch corpus

hooligan 9337

voetbalvandaal 611

success-rate for hooligan: 9337 / (9337 + 611) = 93.9 %

Table 1
These success-rates are calculated semi-automatically for 100 English person reference nouns.
In order to account for variation in these rates, we verify the effect of three conceptual parameters
(lexical field, concept frequency, conceptual newness) and three additional parameters (age of the
loanword, regional variation, donor language frequency). Mixed effect regression shows how the
conceptual features, and most prominently conceptual newness, are most powerful in explaining the
attested variation.
Case Study 2 – Contextual Features: This model does however not account for all variation.
Although denotational synonymy is a prerequisite for the selection of expressions in case study 1,
possible stylistic and pragmatic nuances were not attended to. A contextualized approach to the
issue of lexical choice is thus a necessary addition when trying to fully understand how language
users choose either the anglicism or a (semi-)synonym (Grondelaers & Geeraerts 2003, Geeraerts
2010). This is the purpose of the second case-study, in which we present an in-depth analysis of
junk(ie) vs. verslaafde (“addict”). Relying on previous work (Geeraerts et al. 1994, Divjak & Gries
2006, Edmonds & Hirst 2002), we systematically scrutinize the contexts of use for 11,000 tokens
by means of manual coding. Taking a broad definition of context, we focus on regional preferences
(Belgian vs. Netherlandic Dutch) and register variation (popular vs. quality newspapers). Taking a
more narrow definition of context, we include topic, verify the importance of collocational patterns
(focusing on the object of the addiction for each token – e.g. drugs vs. fashion) and take text
structuring into account (avoidance of repetition). Logistic regression analyses examine the
interplay of all these features in determining lexical choice.

104
Bibliography

Edmonds, P. & G. Hirst. 2002. Near-synonymy and lexical choice. Computational Linguistics 28, 2,
105-144.
Geeraerts, D., S. Grondelaers & P. Bakema. 1994. The Structure of Lexical Variation. Meaning,
Naming and Context. Berlin/New York: Mouton de Gruyter.
Geeraerts, D. 2010. Lexical Variation in Space. In Auer, P. & J. E. Schmidt (eds.), Language and
Space: Theories and Methods, 821-837. Berlin/New York: Mouton de Gruyter.
Gries, S. & D. Divjak. 2006. Ways of trying in Russian: clustering behavioral profiles. Corpus
Linguistics and Linguistic Theory 2, 1, 23-60.
Grondelaers, S. & D. Geeraerts. 2003. Towards a pragmatic model of cognitive onomasiology. In
Cuyckens, H., R. Dirven & J. Taeldeman (eds.), Cognitive Approaches to Lexical
Semantics, 67-92. Berlin/New York: Mouton de Gruyter.
Haspelmath, M. & U. Tadmor (eds.). 2009. Loanwords in the World’s Languages: A Comparative
Handbook. Berlin/New York: Mouton de Gruyter.

105